{"id": "2606.02045", "title": "Attention mechanisms and transfer learning for robust peach leaf damage classification under domain shift", "abstract": "Artificial intelligence provides a practical framework for crop damage assessment from imagery data, supporting early decision-making in agricultural management. In peach orchards, climate change increases abiotic stress and biotic pressures, including pests and diseases, which often produce visually similar foliar symptoms. This overlap makes manual diagnosis difficult, especially across multiple fields with varying environmental conditions, highlighting the need for automated models with strong generalization ability.\n  We propose an image-based classification approach for peach leaf damage detection. A benchmark dataset was created through manual annotation of publicly available images, consisting of 1,366 peach leaves across six damage categories. Several deep learning architectures were evaluated. EfficientNet models achieved the best results, with EfficientNetB0 reaching 92.9 percent accuracy, EfficientNetB3 achieving 91.5 percent, and EfficientNetB5 showing the strongest performance on minority classes. DenseNet121 reached 92.6 percent accuracy. The integration of the Convolutional Block Attention Module (CBAM) improved performance in several backbones, particularly EfficientNetB5 and InceptionV3, while showing limited or negative impact in others. The CBAM-enhanced EfficientNetB5 achieved the best overall accuracy of 93.3 percent.\n  To evaluate robustness under realistic conditions, a local dataset of 180 images across four classes was collected, and transfer learning strategies were applied to address domain shift. Three fine-tuning strategies were tested. EfficientNetB3 combined with CBAM achieved the best performance in the local domain, reaching a 93 percent macro F1-score after transfer. Overall, attention-based models showed improved robustness for minority classes and better generalization across different field conditions.", "date": "2026-06-01", "categories": []}
{"id": "2606.18484", "title": "Vines-DB: An RGB image dataset for multi-species ornamental vine segmentation", "abstract": "The Vines-DB dataset contains 1,218 original high-resolution RGB images of seven ornamental vine species collected under field conditions at the Utah Agricultural Experiment Station's Greenville Research Farm in Logan, Utah, USA. The dataset was generated from 168 individual vine plants that were transplanted in 2022 and photographed repeatedly across multiple months during the 2023 and 2024 growing seasons (July-October). Images were captured with an iPhone 16 Pro equipped with a 48 MP camera between 10:00 AM and 12:00 PM under daylight. Vines were grown on 1.2m x 2.4m trellises and photographed from a distance of 1m against black or white Styrofoam backdrops to improve contrast and reduce background noise. The dataset includes Akebia quinata, Campsis radicans, Hydrangea anomala petiolaris, Lonicera x heckrottii, Campsis x tagliabuana 'Madame Galen', Parthenocissus quinquefolia, and Wisteria floribunda. All original images were manually annotated in Roboflow by trained annotators to produce polygon-based instance segmentation masks for eight classes, including seven species and background. After preprocessing and data augmentation, the working dataset was expanded to 2,307 images for model development and evaluation. The augmented dataset was divided into 2,019 training images, 192 validation images, and 96 test images using stratified sampling to maintain balanced representation. Vines-DB supports the development and evaluation of deep learning models for multi-class instance segmentation in precision horticulture and urban ecology. The dataset enables applications such as automated canopy cover estimation, species identification, and scalable field phenotyping. In addition, repeated monthly imaging of the plants captures temporal variation in canopy development and plant appearance, increasing the dataset's utility for segmentation benchmarking under realistic field conditions.", "date": "2026-06-16", "categories": []}
{"id": "2606.17222", "title": "Quantum Enchanced Multi-Scale CNN with Bi-directional Mamba for Crop Field Analysis", "abstract": "Hyperspectral image (HSI) crop analysis is essential for precision agriculture because it captures rich spectral and spatial information for accurate crop monitoring and assessment. However, HSI classification remains challenging due to high spectral dimensionality, spatial complexity, class imbalance, and limited labeled samples. To address these challenges, this paper proposes a BiSpectral Mamba-based framework that combines multi-scale convolutional feature extraction, spectral attention, bidirectional state-space modeling, and quantum-inspired learning. A multi-scale CNN backbone first extracts hierarchical spatial-spectral representations through feature fusion across multiple resolutions. A spectral attention mechanism then emphasizes informative bands while suppressing redundant and noisy channels. The refined features are processed by a BiSpectral Mamba module that captures long-range dependencies in both forward and backward directions by modeling hyperspectral feature maps as sequential tokens. In addition, class-weighted optimization and feature fusion strategies are incorporated to improve training stability and mitigate class imbalance. Experimental evaluation on the UAVHSI-Crop dataset demonstrates the effectiveness of the proposed framework, achieving an overall accuracy of 84.83%. The results show that integrating convolutional, attention-based, and state-space modeling components enables robust spatial-spectral feature learning for crop classification. The proposed framework also shows potential for broader agricultural and remote sensing applications, including crop disease detection, yield prediction, and soil moisture estimation, while highlighting the effectiveness of structured state-space and quantum-inspired architectures for hyperspectral image analysis.", "date": "2026-06-15", "categories": []}
{"id": "2512.24838", "title": "CropTrack: A Tracking with Re-Identification Framework for Precision Agriculture", "abstract": "Multiple-object tracking (MOT) in agricultural environments presents major challenges due to repetitive patterns, similar object appearances, sudden illumination changes, and frequent occlusions. Contemporary trackers in this domain rely on the motion of objects rather than appearance for association. Nevertheless, they struggle to maintain object identities when targets undergo frequent and strong occlusions. The high similarity of object appearances makes integrating appearance-based association nontrivial for agricultural scenarios. To solve this problem we propose CropTrack, a novel MOT framework based on the combination of appearance and motion information. CropTrack integrates a reranking-enhanced appearance association, a one-to-many association with appearance-based conflict resolution strategy, and an exponential moving average prototype feature bank to improve appearance-based association. Evaluated on publicly available agricultural MOT datasets, CropTrack demonstrates consistent identity preservation, outperforming traditional motion-based tracking methods. Compared to the state of the art, CropTrack achieves significant gains in association accuracy and identification precision scores with a lower number of identity switches.", "date": "2025-12-31", "categories": []}
{"id": "2606.11381", "title": "From Simulation to the Real-World: An In-Field 6D Pose Dataset and Baseline for Robotic Strawberry Harvesting", "abstract": "Robotic strawberry harvesting requires precise 6D pose estimation; however, collecting 6D pose ground truth in real agricultural fields is inherently challenging. Existing strawberry 6D pose estimation studies have therefore relied mainly on synthetic data, often without sufficient scene-level realism,leaving their performance under real agricultural field conditions unquantified. In this work, we present, to the best of our knowledge, the first real-world 6D pose ground truth dataset of strawberries collected in actual agricultural fields (12,040 images). We also introduce a synthetic dataset rendered in NVIDIA Isaac Sim, featuring scene-level realism and domain randomization. Despite this improved simulation setup, our experiments reveal that a substantial sim-to-real gap persists, underscoring the necessity of real agricultural field data for reliable evaluation. We further quantify the sim-to-real gap through baseline 6D pose estimation results across backbone encoders, serving as a reference for future work.", "date": "2026-06-09", "categories": []}
{"id": "2412.14631", "title": "Review of Fruit Tree Image Segmentation", "abstract": "Fruit tree image segmentation is an essential problem in automating a variety of agricultural tasks such as phenotyping, harvesting, spraying, and pruning. Many research papers have proposed a diverse spectrum of solutions suitable to specific tasks and environments. The review scope of this paper is confined to the front views of fruit trees and based on 158 relevant papers collected using a newly designed crawling review method. These papers are systematically reviewed based on a taxonomy that sequentially considers the method, image, task, and fruit. This taxonomy will assist readers to intuitively grasp the big picture of these research activities. Our review reveals that the most noticeable deficiency of the previous studies was the lack of a versatile dataset and segmentation model that could be applied to a variety of tasks and environments. Six important future research tasks are suggested, with the expectation that these will pave the way to building a versatile tree segmentation module.", "date": "2024-12-19", "categories": []}
{"id": "2605.17773", "title": "PlantPose: Universal Plant Skeleton Estimation via Tree-constrained Graph Generation", "abstract": "Accurate estimation of plant skeletal structures (e.g., branching structures) from images is essential for smart agriculture and plant science. Unlike human skeletons with fixed topology, plant skeleton estimation presents a unique challenge, i.e., estimating arbitrary tree graphs from images. To address this problem, we introduce PlantPose, a universal plant skeleton estimator via tree-constrained graph generation. PlantPose combines learning-based graph generation with traditional graph algorithms to enforce tree constraints during the training loop. To enhance the model's generalization capability, we curate a large and diverse dataset comprising real-world and synthetic plant images, along with simplified representations (e.g., sketches and abstract drawings). This dataset enables the generalized model to adapt to diverse input styles and categories of plant images while preserving topological consistency. Our approach demonstrates robust and accurate plant skeleton estimation across multiple domains, including previously unseen out-of-domain scenarios. Further analyses highlight the method's strengths and limitations in handling complex, heterogeneous data distributions. All implementations and datasets are available at https://github.com/huntorochi/PlantPose/.", "date": "2026-05-18", "categories": []}
{"id": "2606.11966", "title": "Feature extraction for plant growth estimation", "abstract": "Precision agriculture requires the estimation of plant growth stages in real-time. When the plant growth stage is known, the wastage of resources in cultivation, such as nutrients and water, is reduced as only the required resources need to be supplied. Plants at different growth stages, however, have similar morphological features, which can make autonomous growth stage estimation difficult. This paper presents two feature extraction methods for growth stage estimation: one that uses a bank of Gabor filters and morphological operations, and the other that uses pre-trained convolutional neural networks (CNNs) and transfer learning. We test these methods on a publicly available plant growth stage dataset (``bccr-segset``) for two species, canola and radish, grown and captured under indoor conditions. The two proposed feature extraction methods are compared, using support vector machines and boosted trees as classifiers. We find that both methods are suitable for real-time applications, and that CNN features outperform the hand-crafted features, both with regard to speed and accuracy. The best system (VGG-19 features, classified with a radial basis function support vector machine) obtained an accuracy of 98.4% for both species, processing an image in 0.08 seconds.", "date": "2026-06-10", "categories": []}
{"id": "2606.06709", "title": "USU-Corn-WeedDB: A UAV RGB Image Dataset for Multi-Species Weed Detection in Forage Corn", "abstract": "Weed pressure in forage corn production causes yield losses of up to 31.5%, yet site-specific weed management (SSWM) systems built on UAV imagery and deep learning remain constrained by the scarcity of field-representative training datasets. We present USU-Corn-WeedDB, a publicly available UAV RGB image dataset collected from a commercial forage corn field in Cache Valley, Utah, designed to support multi-class weed detection under both supervised and semi-supervised learning frameworks. RGB imagery was acquired on 27 June 2025 using an Autel EVO II Dual 640T V2 drone at ~10m above ground level, yielding a ground sampling distance of approximately 0.48 cm/pixel. A total of 366 full-resolution images were tiled into 8,800 patches at 640 x 640-pixel resolution. Of these, 800 images were manually annotated for three weed species; common lambsquarters (Chenopodium album), redroot pigweed (Amaranthus retroflexus), and green foxtail (Setaria viridis) comprising 10,539 bounding-box instances, with the remaining 8,000 tiles retained as an unlabeled pool for semi-supervised experiments. This dataset reflects a natural class imbalance where redroot pigweed constitutes 53.86% of annotated instances, which was preserved intentionally to mirror real field conditions. To validate dataset utility, we trained 28 object detection models spanning five architecture families including YOLOv8, YOLOv9, YOLOv10, YOLO11, YOLO26, and RT-DETR under identical conditions without hyperparameter tuning. Test set mAP@0.5 ranged from 0.773 to 0.840, with lightweight models achieving competitive performance relevant to edge-deployed UAV systems. USU-Corn-WeedDB is publicly available at https://doi.org/10.5281/zenodo.20044178.", "date": "2026-06-04", "categories": []}
{"id": "2606.05611", "title": "What's Under the Skin? Estimating Swine Body Condition", "abstract": "Sow body condition is an important indicator for growers as it has a large impact on lactation performance and piglet survival. However, body condition measures used during production, such as visual scoring and calipers, correlate poorly with underlying tissue composition. Ultrasound scans can provide direct measurements of subcutaneous backfat thickness and loin muscle depth, but their operation is labor intensive and not scalable for production. We present PigFormer, an end-to-end two-stage system that takes raw depth frames from a ceiling-mounted RGB-D camera and predicts subcutaneous backfat thickness, loin muscle depth, and total tissue thickness at the last rib. Stage 1 is a geometric front-end that converts raw depth into a standardized height map via SAM3-to-MaskDINO segmentation distillation, ground-plane removal, and orientation normalization. Stage 2 is a Slice Attention Encoder that treats each height map as a sequence of cross-sectional slices and captures spatial relationships along the full dorsal surface. On a multi-site dataset of 319 sow and gilt instances from two facilities, PigFormer achieves 2.43 mm backfat MAE and 3.87 mm overall MAE. It outperforms strong single-stage ResNet-18 and ViT-small baselines. PigFormer offers a practical path toward continuous, automated, non-contact body condition monitoring in commercial swine production. Code is available at https://github.com/iambashar/Pigformer.", "date": "2026-06-04", "categories": []}
{"id": "2606.06359", "title": "Comparison of Deep Learning Frameworks For Rice Disease Mapping From UAV Multispectral Imaging", "abstract": "In this study, UAV multispectral imagery is used to segment the severity of bacterial leaf blight (BLB) in rice using convolutional neural networks (CNNs) and transformer-based models. The evaluated architectures include U-Net with a ResNet- 101 encoder, U-Net++ with EfficientNet-B3 and EfficientNetB7, DeepLabV3+, and SegFormer, all trained under a common pipeline with three input configurations (multispectral only, multispectral+NDVI, and multispectral+NDRE). Experiments are conducted using the publicly available BLB dataset with performance reported using mean IoU (mIoU), mean F1 (mF1), mean accuracy (mAcc), precision, and recall. U-Net++ with EfficientNet-B3 achieved the highest performance, with an mIoU of 97.62%. SegFormer obtained lower segmentation accuracy but comparable inference speed. Overall, the results indicate that lightweight CNN backbones remain more reliable for operational BLB monitoring while integration of vegetation indices provides small and consistent improvements. The study also highlights the value of standardised UAV datasets to compare disease mapping methods and encourages the use of CNN architectures for field implementation.", "date": "2026-06-04", "categories": []}
{"id": "2411.05359", "title": "Agricultural Landscape Understanding At Country-Scale", "abstract": "Comprehensive agricultural landscape understanding is critical for addressing global challenges in food security, climate change, and resource management. This requires mapping not just crop fields, but also vital features like trees and water bodies which form an intricate mosaic in complex \\textit{smallholder} systems dominating the Global South. Previous efforts to develop such land use maps have been limited by a narrow focus on methods for field delineation only, and also do not develop robust post-processing steps essential for real-world deployment. Further, to our knowledge, no prior system for smallholder farms has been deployed and evaluated at a national scale. This work addresses these limitations by presenting the first national-scale agricultural mapping system that moves beyond simple field delineation to enable segmentation of agricultural instances like fields, trees and water bodies. Our system is refined for real-world application using novel post-processing heuristics to ensure map consistency and accuracy, and is validated through a rigorous, multi-faceted evaluation process. Fine-grained land use maps generated by our system are publicly accessible via an API at \\textit{\\href{http://agri.withgoogle.com}{http://agri.withgoogle.com}}, enabling a wide range of applications from precision agriculture and policy-making to advancing global sustainability development goals.", "date": "2024-11-08", "categories": []}
{"id": "2606.00548", "title": "CAFOSat: A Strongly Annotated Dataset for Infrastructure-Aware CAFO Mapping Using High-Resolution Imagery", "abstract": "Concentrated Animal Feeding Operations (CAFOs) play an important role in agricultural production but are also associated with environmental, public health, and disease surveillance concerns. Large-scale mapping of CAFOs from remote sensing imagery remains challenging due to heterogeneous infrastructure layouts, noisy location records, inconsistent annotations, and incomplete inventories. We introduce CAFOSat, a strongly annotated, infrastructure-aware dataset for CAFO mapping across the United States. CAFOSat integrates high-resolution National Agriculture Imagery Program (NAIP) imagery with multi-source CAFO inventories collected across multiple states and transforms weak geolocation records into refined annotations through a human-in-the-loop pipeline combining AI-assisted annotation, GradCAM-based localization, and geometric clustering. To improve dataset quality, we curate challenging negative samples using land-cover-guided sampling with spatial exclusion constraints and provide infrastructure-level annotations, including barns, manure ponds, and grazing-related features, through manual verification. The resulting dataset contains more than 45,000 image patches spanning 20 states and four major CAFO categories. We benchmark a diverse set of convolutional, transformer-based, and vision-language models, demonstrating the value of refined annotations and curated negative samples for CAFO classification and generalization. In addition, we introduce a synthetic augmentation pipeline that generates infrastructure-aware variations to increase training diversity and improve robustness under distribution shifts. CAFOSat provides a large-scale benchmark for advancing infrastructure-aware agricultural monitoring and CAFO mapping from high-resolution remote sensing imagery.", "date": "2026-05-30", "categories": []}
{"id": "2606.00676", "title": "A Modelling and Evaluation Framework for EuroCrops-Driven Sentinel-2 Crop Segmentation", "abstract": "This work presents a configurable pipeline for generating semantic-segmentation-ready agricultural datasets from Sentinel-2 imagery and EuroCrops parcel-level annotations. The workflow transforms heterogeneous vector crop annotations into aligned multispectral image--mask pairs through label harmonization, Sentinel-2 product selection, spatial alignment, rasterization, patch extraction, quality filtering, and class-aware sample selection. The generated dataset contains 67,337 patches from five European countries and uses a reduced taxonomy of ten crop classes plus background.\n  A four-level U-Net with Group Normalization was trained using 10 Sentinel-2 spectral bands and a composite loss combining class-weighted cross-entropy and Dice loss. On the internal EuroCrops-based test split, the model achieved a mean Intersection over Union (mIoU) of 0.7665, a pixel accuracy of 0.8693, and a mean class accuracy of 0.9072. Compared with spectral and spatial-context Random Forest baselines, the U-Net showed the importance of learned multi-scale spatial representations for crop segmentation.\n  External evaluation was performed on unseen Belgian EuroCrops subsets, DACIA5, and PASTIS. The results show a clear performance gap under external and cross-dataset evaluation, especially for benchmarks with different taxonomies, annotation protocols, spatial coverage, or temporal organization. The model transfers more reliably to dominant and taxonomically aligned classes such as maize and wheat, while performance remains limited for several minority classes and for the adapted single-date PASTIS setting. These findings highlight both the potential and the limitations of using EuroCrops-derived supervision for Sentinel-2 crop segmentation under realistic domain shifts.", "date": "2026-05-30", "categories": []}
{"id": "2605.27129", "title": "YOLO26-RipeLoc Lite: A lightweight architecture for tomato ripeness detection and picking point localization in greenhouse robotic harvesting", "abstract": "In greenhouse tomato production, automated harvesting requires accurate detection of ripe tomatoes, ripeness classification, and precise picking-point localization for robotic end-effectors. This paper proposes YOLO26-RipeLoc Lite, a lightweight deep learning architecture based on YOLO26 for simultaneous detection, ripeness classification, and center-point localization of greenhouse tomatoes. The model introduces three modifications: (1) a Lightweight Feature Pyramid Network (LFPN) with depthwise separable convolutions for efficient multi-scale fusion, (2) a Ripeness-Aware Attention Module (RAAM) with dual pooling and a learnable ripeness bias vector for enhanced color-texture discrimination, and (3) a Compact Detection Head (CDH) with shared convolutions and an integrated center-point regression branch for direct grasp planning. The model is evaluated on a custom dataset of 1,500 images with 6,227 instances (3,566 ripe, 2,661 unripe) from the SILAL greenhouse, Abu Dhabi, UAE. YOLO26-RipeLoc Lite achieves mAP@0.5 of 92.9% (95.2% ripe, 90.6% unripe) with the highest precision (95.2%) among all evaluated architectures using only 2.38M parameters. Post-training BatchNorm pruning at 30% reduces parameters to ~1.8M with negligible accuracy loss. Ablation studies confirm that greenhouse-aware HSV augmentation provides the largest improvement (+2.02 pp mAP@50), backbone freezing achieves peak precision (93.8%), and 3-phase progressive unfreezing yields the best localization quality (mAP@50:95 of 64.6%). Comparisons with YOLOv8n/s, YOLO11n/s, YOLO12n/s, and YOLO26s confirm superior accuracy-efficiency: 2.9 pp higher precision than YOLO12n with 7.0% fewer parameters and integrated center-point localization for robotic end-effector guidance.", "date": "2026-05-26", "categories": []}
{"id": "2605.00908", "title": "Evaluation of Convolutional and Transformer-Based Detectors for Weed Detection in Tomato Plantations", "abstract": "This paper presents a comparative evaluation of convolutional and transformer-based object detection architectures for early weed detection in tomato plantations. Representative models from each paradigm are considered, including YOLOv26-nano, a recent variant of the YOLO family, and RT-DETR Large and RF-DETR Medium as transformer-based architectures. The evaluation was conducted on the GROUNDBASED_WEED dataset, considering six weed classes and an additional category corresponding to unidentified plants, which allowed for the assessment of performance in terms of detection accuracy and computational efficiency using metrics such as precision, recall, average precision, and inference speed, as well as non-parametric statistical tests. The results highlight a clear trade-off between efficiency and contextual modeling: CNN-based detectors achieve high performance at a lower computational cost, while transformer-based approaches offer better global context capture at the expense of higher resource demands. These results provide practical criteria for model selection in precision agriculture applications.", "date": "2026-04-29", "categories": []}
{"id": "2605.24353", "title": "ViViD-5K: Vineyard vision dataset for field-based berry detection and segmentation and grape cluster closure estimation", "abstract": "Cluster closure, defined as the progressive filling of gaps between the berries in a grape bunch, is a key trait in vineyard management, impacting disease risk. However, traditional visual scoring methods are labor-intensive, subjective, and lack temporal resolution. Existing datasets rarely support fine-grained berry-level analysis, limiting the development of robust deep learning models. In this work, we present ViViD-5k, a large-scale in-field Vineyard Vision Dataset containing 5,000 images with dense annotations, including over 648,000 berry centroids and cluster segmentation masks spanning 13 grape varieties. Building on this dataset, we introduce GrapeSAM, a two-stage visual pipeline that combines point-based berry localization with prompt-based segmentation using Segment Anything, followed by transformer-based cluster segmentation. The pipeline enables automated, in-field estimation of cluster closure with minimal supervision. Quantitative results demonstrate strong segmentation and counting accuracy across diverse conditions, while visualizations confirm robustness on both in-domain and out-of-domain samples. This work provides a scalable and objective alternative to manual compactness scoring and supports high-throughput grape phenotyping with enhanced spatial detail.", "date": "2026-05-23", "categories": []}
{"id": "2509.17086", "title": "SFN-YOLO: Towards Free-Range Poultry Detection via Scale-aware Fusion Networks", "abstract": "Detecting and localizing poultry is essential for advancing smart poultry farming. Despite the progress of detection-centric methods, challenges persist in free-range settings due to multiscale targets, obstructions, and complex or dynamic backgrounds. To tackle these challenges, we introduce an innovative poultry detection approach named SFN-YOLO that utilizes scale-aware fusion. This approach combines detailed local features with broader global context to improve detection in intricate environments. Furthermore, we have developed a new expansive dataset (M-SCOPE) tailored for varied free-range conditions. Comprehensive experiments demonstrate our model achieves an mAP of 80.7% with just 7.2M parameters, which is 35.1% fewer than the benchmark, while retaining strong generalization capability across different domains. The efficient and real-time detection capabilities of SFN-YOLO support automated smart poultry farming.", "date": "2025-09-21", "categories": []}
{"id": "2605.22034", "title": "AgroVG: A Large-Scale Multi-Source Benchmark for Agricultural Visual Grounding", "abstract": "Visual grounding, the task of localizing objects described by natural-language expressions, is a foundational capability for agricultural AI systems, enabling applications such as selective weeding, disease monitoring, and targeted harvesting. Reliable evaluation of agricultural visual grounding remains challenging because agricultural targets are often small, repetitive, occluded, or irregularly shaped, and instructions may refer to one, many, or no objects in an image. Evaluating this capability therefore requires jointly testing localization accuracy, target-set completeness, and existence-aware abstention. To address these challenges, we introduce \\textbf{AgroVG}, a multi-source benchmark that formulates agricultural grounding as generalized set prediction: given an image and a referring expression, a model must return all matching target instances or abstain when no target is present. AgroVG contains 10{,}071 annotation-grounded image-query pairs from ten source datasets across six target families: crop/weed, fruit, wheat head, pest, plant disease, and tree canopy. It supports bounding-box grounding (T1) across all six families and instance-mask grounding (T2) on sources with reliable instance-level pixel annotations, with queries covering single-target, multi-target, and target-absent regimes. AgroVG further provides task-specific protocols for box-set matching and query-level mask coverage. Zero-shot evaluation of 26 model configurations spanning closed-source MLLMs, open-source VLMs, and specialized grounding systems reveals persistent gaps: the best multi-target Set-$F_1$ reaches only 0.35, and the best positive-query mask success rate at IoU@0.75 remains below 0.17. Data and code are available at https://anonymous.4open.science/r/AgroVG-5172/ .", "date": "2026-05-21", "categories": []}
{"id": "2412.01944", "title": "A Comparative Study of Transformer and Convolutional Models for Crop Segmentation from Satellite Image Time Series", "abstract": "Crop segmentation from satellite image time series (SITS) is a fundamental task for agricultural monitoring and land-use analysis. While convolutional neural networks (CNNs) have been widely used, transformer-based architectures offer alternative mechanisms for representing spatial and temporal dependencies in multispectral data. This paper presents a comparative study of CNN and transformer-based segmentation models for crop mapping from Sentinel-2 time series, including 3D U-Net, 3D FPN, 3D DeepLabv3, and three transformer architectures: Swin UNETR, TSViT, and VistaFormer, which adopt different strategies for capturing temporal dependencies. Experiments on the Munich and Lombardia datasets show that TSViT achieves the best overall results, slightly surpassing 3D U-Net, which remains a strong CNN baseline. VistaFormer offers the best efficiency, while Swin UNETR performs competitively but is less effective than transformers that explicitly model temporal dynamics. These results highlight that temporal modelling is critical for SITS: TSViT outperforms CNNs and approaches that treat time as an additional spatial dimension, while VistaFormer provides a strong efficiency-performance trade-off.", "date": "2024-12-02", "categories": []}
{"id": "2605.20940", "title": "3D Reconstruction and Knowledge Distillation to Improve Multi-View Image Models to Explore Spike Volume Estimation in Wheat", "abstract": "Accurate estimation of wheat spike volume is important for yield component analysis and stress resilience assessment, yet field-based measurement remains challenging. Active 3D sensing methods such as Light Detection and Ranging (LiDAR) or time-of-flight (ToF) are sensitive to plant motion or poorly suited to outdoor conditions, while 3D reconstructions are computationally expensive. Direct 2D image processing would offer computational advantages, but image-based models lack explicit geometric information. We therefore propose a hybrid 2D-3D approach with knowledge distillation during training while enabling efficient image-only inference. First, we train a rigid-invariant point cloud network using distance-based histogram features to obtain pose-robust geometric representations. We then combine the 3D model with a proposed multi-view image-based regulated Transformer (RT) in an ensemble architecture. Finally, we distill the ensemble knowledge into a purely image-based student model using either feature-based or label-based distillation. The two distilled RTs reduce the mean absolute error (MAE) from 654.31 mm$^3$ of the non-distilled RT to 639.93 mm$^3$ and 644.62 mm$^3$, and increase correlation from 0.76 to 0.77 and 0.82, respectively. At the same time, inference time is reduced from 160 ms to 1.4 ms per spike. Distillation further mitigates volume-dependent bias and reshapes the latent representation of the image model toward a geometry-aware shape. Our results demonstrate that 3D-informed training of a 2D Transformer allows for scalable and efficient spike volume estimation for high-throughput field phenotyping.", "date": "2026-05-20", "categories": []}
{"id": "2605.11055", "title": "The first global agricultural field boundary map at 10m resolution", "abstract": "The agricultural field is the natural unit at which crops are planted, managed, regulated, and reported, yet most global remote-sensing products for agriculture are only available at the pixel level. While some high-quality field-level data products exist, they come from parcel registries covering only parts of Europe or from ML-derived products for individual countries. No openly available, globally consistent map of agricultural field boundaries exists to date. Here we present the first global field boundary dataset at 10\\,m resolution for the years 2024 and 2025, comprising 3.17 billion remote-sensing field polygons (1.62 B in 2024 and 1.55 B in 2025) across 241 countries and territories, produced by applying a U-Net segmentation model trained on the Fields of The World dataset to cloud-free Sentinel-2 mosaics. Validated against ground-truth field boundaries in 24 countries, the map achieved a mean pixel-level recall of 0.85 with 14 countries exceeding 0.90. Evaluation against full-country ground-truth datasets in Austria, Latvia, and Finland yielded F1 scores of 0.89, 0.88, and 0.74, respectively. Because reference data for global validation is inherently incomplete, we accompanied the map with a 500 m confidence layer that identifies regions where predictions are reliable. We release the dataset openly as three global maps: the confidence-thresholded default field boundary dataset, the full unfiltered dataset, and the continuous-valued confidence raster. These maps provide the first globally consistent field-level unit of analysis for crop monitoring, food security, and downstream agricultural science.", "date": "2026-05-11", "categories": []}
{"id": "2605.12282", "title": "Large-Small Model Collaboration for Farmland Semantic Change Detection", "abstract": "Farmland Semantic Change Detection (SCD) is essential for cultivated land protection, yet existing benchmarks and models remain insufficient for fine-grained farmland conversion monitoring. Current datasets often lack dedicated \"from-to\" annotations, while visual change detection models are easily disturbed by phenology-induced pseudo-changes caused by crop rotation, seasonal variation, and illumination differences. To address these challenges, we construct HZNU-FCD, a large-scale fine-grained farmland SCD benchmark with a unified five-class farmland-to-non-farmland annotation protocol. It contains 4,588 bitemporal image pairs with pixel-level labels for practical farmland protection. Based on this benchmark, we propose a large-small collaborative SCD framework that integrates a task-driven small visual model with a frozen large vision-language model. The small model, Fine-grained Difference-aware Mamba (FD-Mamba), learns dense change representations for boundary preservation and small-region localization. The large-model pathway, Cross-modal Logical Arbitration (CMLA), introduces CLIP-based textual priors for prompt-guided semantic arbitration and pseudo-change suppression. To enable effective collaboration, we design a hard-region co-training strategy that supervises the CMLA semantic score map only on low-confidence pixels. Experiments show that our method achieves 97.63% F1, 96.32% IoU, and 96.35% SCD_IoU_mean on HZNU-FCD with only 6.65M trainable parameters. Compared with the multimodal ChangeCLIP-ViT, which leverages vision-language information for change detection, our method improves F1 by 10.19 percentage points on HZNU-FCD. It also achieves 91.43% F1 and 84.21% IoU on LEVIR-CD, and 93.85% F1 and 88.41% IoU on WHU-CD, demonstrating strong robustness and generalization. The code is available at https://github.com/Lovelymili/FD-Mamba.", "date": "2026-05-12", "categories": []}
{"id": "2605.08213", "title": "Low-Cost Stereo Vision for Robust 3D Positioning of Thin Radiata Pine Branches in Autonomous Drone Pruning", "abstract": "Manual pruning of radiata pine, a species of major economic importance to New Zealand forestry, is hazardous, labour-intensive, and increasingly constrained by workforce shortages. Existing autonomous pruning platforms typically rely on expensive sensors such as LiDAR and are limited to thick branches, which restricts their wider adoption. This paper investigates whether a single low-cost stereo camera mounted on a drone can provide sufficiently accurate branch detection and three-dimensional positioning to support autonomous pruning of branches as thin as 10 mm, thereby removing the need for auxiliary depth sensors. The proposed pipeline comprises two stages: branch segmentation and depth estimation. For segmentation, Mask R-CNN variants and the YOLOv8 and YOLOv9 families are compared on a custom dataset of 71 stereo image pairs captured with a ZED Mini camera; YOLOv8 and YOLOv9 are selected as representative state-of-the-art real-time segmentors at the time of data collection, and the framework is designed to remain compatible with newer YOLO releases. For depth estimation, a traditional method (SGBM with WLS filtering) and deep-learning-based methods (PSMNet, ACVNet, GWCNet, MobileStereoNet, RAFT-Stereo, and NeRF-Supervised Deep Stereo) are evaluated, including cross-dataset fine-tuning experiments that expose the domain gap between urban driving benchmarks and natural forestry scenes. The main novelty of this work lies in coupling stereo segmentation with a centroid-based triangulation algorithm and Median-Absolute-Deviation outlier rejection that converts a segmentation mask and disparity map into a single robust branch-to-camera distance, addressing the challenges of sparse texture, thin structures, and noisy disparity values typical of forest scenes. Qualitative evaluations at distances of 1-2 m show that the learning-based stereo methods produce more coherent depth es...", "date": "2026-05-06", "categories": []}
{"id": "2509.12047", "title": "A Computer Vision Pipeline for Individual-Level Behavior Analysis: Benchmarking on the Edinburgh Pig Dataset", "abstract": "Animal behavior analysis plays a crucial role in understanding animal welfare, health status, and productivity in agricultural settings. However, traditional manual observation methods are time-consuming, subjective, and limited in scalability. We present a modular pipeline that leverages open-sourced state-of-the-art computer vision techniques to automate animal behavior analysis in a group housing environment. Our approach combines state-of-the-art models for zero-shot object detection, motion-aware segmentation and tracking, and advanced feature extraction using vision transformers for robust behavior recognition. The pipeline addresses challenges including animal occlusions and group housing scenarios, as demonstrated in indoor pig monitoring. We validated our system on the Edinburgh Pig Behavior Video Dataset for multiple behavioral tasks. Our temporal model achieved 94.2% overall accuracy, representing a 21.2 percentage point improvement over existing methods. The pipeline demonstrated robust tracking capabilities with a 93.3% identity preservation (IDF1) score and an 89.3% average precision (AP) for object detection. The modular design suggests potential for adaptation to other contexts, though further validation across species would be required. The open-source implementation provides a scalable solution for behavior monitoring, contributing to precision pig farming and welfare assessment through automated, objective, and continuous analysis.", "date": "2025-09-15", "categories": []}
{"id": "2603.07819", "title": "Fusion Complexity Inversion: Why Simpler Cross View Modules Outperform SSMs and Cross View Attention Transformers for Pasture Biomass Regression", "abstract": "Accurate estimation of pasture biomass from agricultural imagery is critical for sustainable livestock management, yet existing methods are limited by the small, imbalanced, and sparsely annotated datasets typical of real world monitoring. In this study, adaptation of vision foundation models to agricultural regression is systematically evaluated on the CSIRO Pasture Biomass benchmark, a 357 image dual view dataset with laboratory validated, component wise ground truth for five biomass targets, through 17 configurations spanning four backbones (EfficientNet-B3 to DINOv3-ViT-L), five cross view fusion mechanisms, and a 4x2 metadata factorial. A counterintuitive principle, termed \"fusion complexity inversion\", is uncovered: on scarce agricultural data, a two layer gated depthwise convolution (R^2 = 0.903) outperforms cross view attention transformers (0.833), bidirectional SSMs (0.819), and full Mamba (0.793, below the no fusion baseline). Backbone pretraining scale is found to monotonically dominate all architectural choices, with the DINOv2 -> DINOv3 upgrade alone yielding +5.0 R^2 points. Training only metadata (species, state, and NDVI) is shown to create a universal ceiling at R^2 ~ 0.829, collapsing an 8.4 point fusion spread to 0.1 points. Actionable guidelines for sparse agricultural benchmarks are established: backbone quality should be prioritized over fusion complexity, local modules preferred over global alternatives, and features unavailable at inference excluded.", "date": "2026-03-08", "categories": []}
{"id": "2605.01355", "title": "AgriKD: Cross-Architecture Knowledge Distillation for Efficient Leaf Disease Classification", "abstract": "Automated leaf disease classification is critical for early disease detection in resource-constrained field environments. Vision Transformers (ViTs) provide strong representation capability by modeling long-range dependencies and inter-class relationships; however, their high computational cost makes them impractical for deployment on edge devices. As a result, existing approaches struggle to effectively transfer these rich representations to lightweight models. This paper introduces AgriKD, a cross-architecture knowledge distillation framework for efficient edge deployment, which transfers knowledge from a Vision Transformer (ViT) teacher to a compact convolutional student model. To bridge the representational gap between Transformer and CNN architectures, the proposed approach integrates multiple distillation objectives at the output, feature, and relational levels, where each objective captures a different aspect of the teacher knowledge. This enables the student model to better preserve and utilize transformer-derived global representations. Experiments on multiple leaf disease datasets show that the distilled student achieves performance comparable to the teacher while significantly improving efficiency, reducing model parameters by approximately 172 times, computational cost by 47.57 times, and inference latency by 18-22 times. Furthermore, the optimized model is deployed across multiple runtime formats, including ONNX, TFLite Float16, and TensorRT FP16, achieving consistent predictive performance with negligible accuracy degradation. Real-world deployment on NVIDIA Jetson edge devices and a mobile application demonstrates reliable real-time inference, highlighting the practicality of AgriKD for AI-powered agricultural applications in resource-constrained environments.", "date": "2026-05-02", "categories": []}
{"id": "2605.06121", "title": "Pest-Thinker: Learning to Think and Reason like Entomologists via Reinforcement Learning", "abstract": "Pest-induced crop losses pose a major threat to global food security and sustainable agricultural development. While recent advances in Multimodal Large Language Models (MLLMs) have shown strong potential for visual understanding and smart agriculture, their direct application to pest recognition remains limited due to the domain's unique challenges such as high inter-species complexity, intra-species variability, and the scarcity of expert-annotated data. In this work, we introduce Pest-Thinker, a knowledge-driven reinforcement learning (RL) framework that enables MLLMs to reason over fine-grained pest morphology. We first construct two high-definition pest benchmarks, QFSD and AgriInsect, comprising diverse species and expert-annotated morphological traits. Leveraging these datasets, we synthesize Chain-of-Thought (CoT) reasoning trajectories to facilitate structured learning of pest-specific visual cues through Supervised Fine-Tuning (SFT). Subsequently, we employ Group Relative Policy Optimization (GRPO) with a novel feature reward that guides the model to focus on observable morphological evidence, assessed by an LLM-as-a-Judge strategy. Extensive experiments demonstrate that Pest-Thinker substantially improves both in-domain and out-of-domain morphological understanding, marking a step toward expert-level visual reasoning for intelligent agricultural pest analysis. The datasets and source code are available upon acceptance.", "date": "2026-05-07", "categories": []}
{"id": "2504.16684", "title": "SemanticSugarBeets: A Multi-Task Framework and Dataset for Inspecting Harvest and Storage Characteristics of Sugar Beets", "abstract": "While sugar beets are stored prior to processing, they lose sugar due to factors such as microorganisms present in adherent soil and excess vegetation. Their automated visual inspection promises to aide in quality assurance and thereby increase efficiency throughout the processing chain of sugar production. In this work, we present a novel high-quality annotated dataset and two-stage method for the detection, semantic segmentation and mass estimation of post-harvest and post-storage sugar beets in monocular RGB images. We conduct extensive ablation experiments for the detection of sugar beets and their fine-grained semantic segmentation regarding damages, rot, soil adhesion and excess vegetation. For these tasks, we evaluate multiple image sizes, model architectures and encoders, as well as the influence of environmental conditions. Our experiments show an mAP50-95 of 98.8 for sugar-beet detection and an mIoU of 64.0 for the best-performing segmentation model.", "date": "2025-04-23", "categories": []}
{"id": "2507.22101", "title": "AI in Agriculture: A Survey of Deep Learning Techniques for Crops, Fisheries and Livestock", "abstract": "Crops, fisheries and livestock form the backbone of global food production, essential to feed the ever-growing global population. However, these sectors face considerable challenges, including climate variability, resource limitations, and the need for sustainable management. Addressing these issues requires efficient, accurate, and scalable technological solutions, highlighting the importance of artificial intelligence (AI). This survey presents a systematic and thorough review of more than 200 research works covering conventional machine learning approaches, advanced deep learning techniques (e.g., vision transformers), and recent vision-language foundation models (e.g., CLIP) in the agriculture domain, focusing on diverse tasks such as crop disease detection, livestock health management, and aquatic species monitoring. We further cover major implementation challenges such as data variability and experimental aspects: datasets, performance evaluation metrics, and geographical focus. We finish the survey by discussing potential open research directions emphasizing the need for multimodal data integration, efficient edge-device deployment, and domain-adaptable AI models for diverse farming environments. Rapid growth of evolving developments in this field can be actively tracked on our project page: https://github.com/umair1221/AI-in-Agriculture", "date": "2025-07-29", "categories": []}
{"id": "2605.00903", "title": "A Light Weight Multi-Features-View Convolution Neural Network For Plant Disease Identification", "abstract": "Agriculture is a key sector of the economies of developing countries. It serves as a primary source of income and employment for rural populations. However, each year, a large portion of crops is wasted because of pests and diseases. Well-timed prediction of plant diseases is crucial to sustainable, high-quality agricultural production. Detection of plant diseases through conventional methods is both labour-intensive and time-consuming. Researchers have developed image classification based automated techniques for this purpose. Most accurate methods are based on deep convolutional neural networks, which are computationally intensive, with many layers and millions of trainable parameters. In resource-constrained settings, especially in rural areas, it is difficult to deploy deep convolutional neural network models for efficient plant disease identification. To address these issues, an efficient and light-weight Multi-View Convolutional Neural Network is proposed. These additional features aid the proposed model to identify the plant diseases accurately and efficiently with less number of parameters. The proposed model is tested on a benchmark Plantvillage dataset and achieves an improvement of $ 2.9\\%$ in classification accuracy compared to the baseline convolutional neural network model, which was trained only on Red, Green, and Blue (RGB) plant images. Compared with state-of-the-art deep convolutional neural network models, the proposed model is less computationally expensive and achieves comparable accuracy for plant disease identification on the PlantVillage dataset.", "date": "2026-04-28", "categories": []}
{"id": "2604.26084", "title": "FruitProM-V2: Robust Probabilistic Maturity Estimation and Detection of Fruits and Vegetables", "abstract": "Accurate fruit maturity identification is essential for determining harvest timing, as incorrect assessment directly affects yield and post-harvest quality. Although ripening is a continuous biological process, vision-based maturity estimation is typically formulated as a multi-class classification task, which imposes sharp boundaries between visually similar stages. To examine this limitation, we perform an annotation reliability study with two independent annotators on a held-out tomato dataset and observe disagreement concentrated near adjacent maturity stages. Motivated by this observation, we model maturity as a latent continuous variable and predict it probabilistically using a distributional detection head, converting the distribution into class probabilities through the cumulative distribution function (CDF). The proposed formulation maintains comparable performance to a standard detector under clean labels while better representing uncertainty. Furthermore, when controlled label noise is introduced during training, the probabilistic model demonstrates improved robustness relative to the baseline, indicating that explicitly modeling maturity uncertainty leads to more reliable visual maturity estimation.", "date": "2026-04-28", "categories": []}
{"id": "2604.22788", "title": "Non-Destructive Prediction of Fruit Ripeness and Firmness Using Hyperspectral Imaging and Lightweight Machine Learning Models", "abstract": "Post-harvest fruit quality assessment is essential for reducing food waste, yet reliable non-destructive methods typically depend on expensive hyperspectral cameras and computationally intensive deep learning models. These systems typically require GPU resources, large-scale training data, and domain expertise, limiting their feasibility for many real-world agricultural settings. This study systematically evaluates 20 classical machine learning algorithms on hyperspectral imaging data for simultaneous ripeness classification and firmness prediction across five fruit species, using cross-validated experimental design with Bayesian hyperparameter optimization. Data preprocessing strategy, particularly class balancing and spectral transformations, contributes as much to prediction accuracy as algorithm choice. Our results show that tree-based machine learning models can outperform state-of-the-art deep earning models reported in Fruit-HSNet. Moreover, the findings indicate that only three visible-range wavelengths are needed to recover over 94% of full-spectrum accuracy, demonstrating that low-cost multispectral sensors combined with lightweight machine learning models can serve as practical alternatives to expensive hyperspectral cameras and complex deep learning approaches for practical fruit quality sorting.", "date": "2026-04-05", "categories": []}
{"id": "2604.23653", "title": "ResAF-Net: An Anchor-Free Attention-Based Network for Tree Detection and Agricultural Mapping in Palestine", "abstract": "Reliable agricultural data is essential for food security, land-use planning, and economic resilience, yet in Palestine, such data remains difficult to collect at scale because of fragmented landscapes, limited field access, and restrictions on aerial monitoring. This paper presents ResAF-Net, a satellite-based tree detection framework designed for large-scale agricultural monitoring in resource-constrained settings. The proposed architecture combines a ResNet-50 encoder, Atrous Spatial Pyramid Pooling (ASPP), a feature-fusion stage, a multi-head self-attention refinement module, and an anchor-free FCOS detection head to improve tree localization in dense and heterogeneous scenes. Trained on the MillionTrees benchmark, the model achieved 82% Recall, 63.03% mAP@0.50, and 35.47% mAP@0.50:0.95 on the validation split, indicating strong sensitivity to tree presence while maintaining competitive localization quality. Beyond benchmark evaluation, we implemented the model within a web-based GIS application integrated with Palestinian cadastral data from GeoMolg, enabling tree analysis at scene, parcel, and community levels. This deployment demonstrates the practical feasibility of AI-assisted agricultural inventorying in Palestine. It provides a foundation for data-driven monitoring, reporting, and future species-level analysis of Mediterranean tree crops.", "date": "2026-04-26", "categories": []}
{"id": "2604.23701", "title": "Agri-CPJ: A Training-Free Explainable Framework for Agricultural Pest Diagnosis Using Caption-Prompt-Judge and LLM-as-a-Judge", "abstract": "Crop disease diagnosis from field photographs faces two recurring problems: models that score well on benchmarks frequently hallucinate species names, and when predictions are correct, the reasoning behind them is typically inaccessible to the practitioner. This paper describes Agri-CPJ (Caption-Prompt-Judge), a training-free few-shot framework in which a large vision-language model first generates a structured morphological caption, iteratively refined through multi-dimensional quality gating, before any diagnostic question is answered. Two candidate responses are then generated from complementary viewpoints, and an LLM judge selects the stronger one based on domain-specific criteria. Caption refinement is the component with the largest individual impact: ablations confirm that skipping it consistently degrades downstream accuracy across both models tested. On CDDMBench, pairing GPT-5-Nano with GPT-5-mini-generated captions yields \\textbf{+22.7} pp in disease classification and \\textbf{+19.5} points in QA score over no-caption baselines. Evaluated without modification on AgMMU-MCQs, GPT-5-Nano reached 77.84\\% and Qwen-VL-Chat reached 64.54\\%, placing them at or above most open-source models of comparable scale despite the format shift from open-ended to multiple-choice. The structured caption and judge rationale together constitute a readable audit trail: a practitioner who disagrees with a diagnosis can identify the specific caption observation that was incorrect. Code and data are publicly available https://github.com/CPJ-Agricultural/CPJ-Agricultural-Diagnosis", "date": "2026-04-26", "categories": []}
{"id": "2604.23776", "title": "From Noisy Historical Maps to Time-Series Oil Palm Mapping Without Annotation in Malaysia and Indonesia (2020-2024)", "abstract": "Accurate monitoring of oil palm plantations is critical for balancing economic development with environmental conservation in Southeast Asia. However, existing plantation maps often suffer from low spatial resolution and a lack of recent temporal coverage, impeding effective surveillance of rapid land-use changes. In this study, we propose a deep learning framework to generate 10-meter resolution oil palm plantation maps for Indonesia and Malaysia from 2020 to 2024, utilizing Sentinel-2 imagery without requiring new manual annotations. To address the resolution mismatch between coarse 100-meter historical labels and 10-meter imagery, we employ a U-Net architecture optimized with Determinant-based Mutual Information (DMI). This approach effectively mitigates the influence of label noise. We validated our method against 2,058 manually verified points, achieving overall accuracies of 70.64%, 63.53%, and 60.06% for the years 2020, 2022, and 2024, respectively. Our comprehensive analysis reveals that oil palm coverage in the region peaked in 2022 before experiencing a decline in 2024. Furthermore, land cover transition analysis highlights a concerning trajectory of plantation expansion into flooded vegetation areas, despite a general stabilization in rotations with other crop types. These high-resolution maps provide essential data for monitoring sustainability commitments and deforestation dynamics in the region, and the generated datasets are made publicly available at https://doi.org/10.5281/zenodo.17768444.", "date": "2026-04-26", "categories": []}
{"id": "2503.01605", "title": "A Leaf-Level Dataset for Soybean-Cotton Detection and Segmentation", "abstract": "Soybean and cotton are major drivers of many countries' agricultural sectors, offering substantial economic returns but also facing persistent challenges from volunteer plants and weeds that hamper sustainable management. Effectively controlling volunteer plants and weeds demands advanced recognition strategies that can identify these amidst complex crop canopies. While deep learning methods have demonstrated promising results for leaf-level detection and segmentation, existing datasets often fail to capture the complexity of real-world agricultural fields. To address this, we collected 640 high-resolution images from a commercial farm spanning multiple growth stages, weed pressures, and lighting variations. Each image is annotated at the leaf-instance level, with 7,221 soybean and 5,190 cotton leaves labeled via bounding boxes and segmentation masks, capturing overlapping foliage, small leaf size, and morphological similarities. We validate this dataset using YOLOv11, demonstrating state-of-the-art performance in accurately identifying and segmenting overlapping foliage. Our publicly available dataset supports advanced applications such as selective herbicide spraying and pest monitoring and can foster more robust, data-driven strategies for soybean-cotton management.", "date": "2025-03-03", "categories": []}
{"id": "2604.18632", "title": "StomaD2: An All-in-One System for Intelligent Stomatal Phenotype Analysis via Diffusion-Based Restoration Detection Network", "abstract": "Stomata play a crucial role in regulating plant physiological processes and reflecting environmental responses. However, accurate and high-throughput stomatal phenotyping remains challenging, as conventional approaches rely on destructive sampling and manual annotation, restricting large-scale and field deployment. To overcome these limitations, a noninvasive restoration-detection integrated framework, termed StomaD2, is developed to achieve accurate and fast stomatal phenotyping under complex imaging conditions. The framework incorporates a diffusion-based restoration module to recover degraded images and a specialized rotated object detection network tailored to the small, dense, and cluttered characteristics of stomata. The proposed network enhances feature representation through three key innovations: a column-wise structure for global feature interaction, context-aware resampling and reweighting mechanism to improve multi-scale consistency, and a feature reassembly module to boost discrimination against complex backgrounds. In extensive comparisons, StomaD2 demonstrated state-of-the-art performance. On public Maize and Wheat datasets, it achieved accuracies of 0.994 and 0.992, respectively, significantly outperforming existing benchmarks. When benchmarked against ten other advanced models, including Oriented Former and YOLOv12, StomaD2 achieved a top-tier F1-score/mAP of 0.989. The framework is integrated into a user-friendly, field-operable system that supports the fast extraction of eight stomatal phenotypes, such as density and conductance. Validated on more than 130 plant species, StomaD2's results highlight its strong generalizability and potential for large-scale phenotyping, plant physiology analysis, and precision agriculture applications.", "date": "2026-04-18", "categories": []}
{"id": "2507.17869", "title": "Integrating Feature Selection and Machine Learning for Nitrogen Assessment in Grapevine Leaves using In-Field Hyperspectral Imaging", "abstract": "Nitrogen (N) is one of the most critical nutrients in winegrape production, influencing vine vigor, fruit composition, and wine quality. Because soil N availability varies spatially and temporally, accurate estimation of leaf N concentration is essential for optimizing fertilization at the individual plant level. In this study, in-field hyperspectral images (400-1000 nm) were collected from four grapevine cultivars (Chardonnay, Pinot Noir, Concord, and Syrah) across two growth stages (bloom and veraison) during the 2022 and 2023 growing seasons at both the leaf and canopy levels. An ensemble feature selection framework was developed to identify the most informative spectral bands for N estimation within individual cultivars, effectively reducing redundancy and selecting compact, physiologically meaningful band combinations spanning the visible, red-edge, and near-infrared regions. At the leaf level, models achieved the highest predictive accuracy for Chardonnay (R^2 = 0.82, RMSE = 0.19 %DW) and Pinot Noir (R^2 = 0.69, RMSE = 0.20 %DW). Canopy-level predictions also performed well, with R^2 values of 0.65, 0.72, and 0.70 for Chardonnay, Concord, and Syrah, respectively. White cultivars exhibited balanced spectral contributions across the visible, red-edge, and near-infrared regions, whereas red cultivars relied more heavily on visible bands due to anthocyanin-chlorophyll interactions. Leaf-level N-sensitive bands selected for Chardonnay and Pinot Noir were successfully transferred to the canopy level, improving or maintaining prediction accuracy across cultivars. These results confirm that ensemble feature selection captures spectrally robust, scale-consistent bands transferable across measurement levels and cultivars, demonstrating the potential of integrating in-field hyperspectral imaging with machine learning for vineyard N status monitoring.", "date": "2025-07-23", "categories": []}
{"id": "2604.16480", "title": "Positioning radiata pine branches requiring pruning by drone stereo vision", "abstract": "This paper presents a stereo-vision-based system mounted on a drone for detecting and localising radiata pine branches to support autonomous pruning. The proposed pipeline comprises two stages: branch segmentation and depth estimation. For segmentation, YOLOv8, YOLOv9, and Mask R-CNN variants are compared on a custom dataset of 71 stereo image pairs captured with a ZED Mini camera. For depth estimation, both a traditional method (SGBM with WLS filtering) and deep-learning-based methods (PSMNet, ACVNet, GWCNet, MobileStereoNet, RAFT-Stereo, and NeRF-Supervised Deep Stereo) are evaluated. A centroid-based triangulation algorithm with MAD outlier rejection is proposed to compute branch distance from the segmentation mask and disparity map. Qualitative evaluation at distances of 1-2 m indicates that the deep learning-based disparity maps produce more coherent depth estimates than SGBM, demonstrating the feasibility of low-cost stereo vision for automated branch positioning in forestry.", "date": "2026-04-12", "categories": []}
{"id": "2511.13417", "title": "Delineate Anything Flow: Fast, Country-Level Field Boundary Detection from Any Source", "abstract": "Accurate delineation of agricultural field boundaries from satellite imagery is essential for land management and crop monitoring, yet existing methods often produce incomplete boundaries, merge adjacent fields, and struggle to scale. We present the Delineate Anything Flow (DelAnyFlow) methodology, a resolution-agnostic approach for large-scale field boundary mapping. DelAnyFlow combines the DelAny instance segmentation model, based on a YOLOv11 backbone and trained on the large-scale Field Boundary Instance Segmentation-22M (FBIS 22M) dataset, with a structured post-processing, merging, vectorization, and simplification to generate vector boundaries. FBIS 22M, the largest dataset of its kind, contains 672,909 multi-resolution image patches (0.25-10m) and 22.9million validated field instances. The DelAny model delivers state-of-the-art accuracy with over 100% higher mAP and 400x faster inference than SAM2. DelAny demonstrates strong zero-shot generalization and supports national-scale applications: using Sentinel 2 data for 2024, DelAnyFlow generated a complete field boundary layer for Ukraine (603,000km2) in under six hours on a single workstation. DelAnyFlow outputs significantly improve boundary completeness relative to operational products from Sinergise Solutions and NASA Harvest, particularly in smallholder and fragmented systems (0.25-1ha). For Ukraine, DelAnyFlow delineated 3.75M fields at 5m and 5.15M at 2.5m, compared to 2.66M detected by Sinergise Solutions and 1.69M by NASA Harvest. This work delivers a scalable, cost-effective methodology for field delineation in regions lacking digital cadastral data. A project landing page with links to model weights, code, national-scale vector outputs, and dataset is available at https://lavreniuk.github.io/Delineate-Anything/.", "date": "2025-11-17", "categories": []}
{"id": "2509.03497", "title": "Invariant Features for Global Crop Type Classification", "abstract": "Accurate global crop type mapping supports agricultural monitoring and food security, yet remains limited by the scarcity of labeled data in many regions. A key challenge is enabling models trained in one geography to generalize reliably to others despite shifts in climate, phenology, and spectral characteristics. In this work, we show that geographic transfer in crop classification is primarily governed by the ability to learn invariant structure in multispectral time series. To systematically study this, we introduce CropGlobe, a globally distributed benchmark dataset of 300,000 samples spanning eight countries and five continents, and define progressively harder transfer settings from cross-country to cross-hemisphere. Across all settings, we find that simple spectral-temporal representations outperform both handcrafted features and modern geospatial foundation model embeddings. We propose CropNet, a lightweight convolutional architecture that jointly convolves across spectral and temporal dimensions to learn invariant crop signatures. Despite its simplicity, CropNet consistently outperforms larger transformer-based and foundation-model approaches under geographic domain shift. To further improve robustness to geographic variation, we introduce augmentations that simulate shifts in crop phenology and reflectance. Combined with CropNet, this yields substantial gains under large domain shifts. Our results demonstrate that inductive bias toward joint spectral-temporal structure is more critical for transfer than model scale or pretraining, pointing toward a scalable and data-efficient paradigm for worldwide agricultural mapping. Data and code are available at https://github.com/x-ytong/CropNet/.", "date": "2025-09-03", "categories": []}
{"id": "2604.09920", "title": "Does Your VFM Speak Plant? The Botanical Grammar of Vision Foundation Models for Object Detection", "abstract": "Vision foundation models (VFMs) offer the promise of zero-shot object detection without task-specific training data, yet their performance in complex agricultural scenes remains highly sensitive to text prompt construction. We present a systematic prompt optimization framework evaluating four open-vocabulary detectors -- YOLO World, SAM3, Grounding DINO, and OWLv2 -- for cowpea flower and pod detection across synthetic and real field imagery. We decompose prompts into eight axes and conduct one-factor-at-a-time analysis followed by combinatorial optimization, revealing that models respond divergently to prompt structure: conditions that optimize one architecture can collapse another. Applying model-specific combinatorial prompts yields substantial gains over a naive species-name baseline, including +0.357 mAP@0.5 for YOLO World and +0.362 mAP@0.5 for OWLv2 on synthetic cowpea flower data. To evaluate cross-task generalization, we use an LLM to translate the discovered axis structure to a morphologically distinct target -- cowpea pods -- and compare against prompting using the discovered optimal structures from synthetic flower data. Crucially, prompt structures optimized exclusively on synthetic data transfer effectively to real-world fields: synthetic-pipeline prompts match or exceed those discovered on labeled real data for the majority of model-object combinations (flower: 0.374 vs. 0.353 for YOLO World; pod: 0.429 vs. 0.371 for SAM3). Our findings demonstrate that prompt engineering can substantially close the gap between zero-shot VFMs and supervised detectors without requiring manual annotation, and that optimal prompts are model-specific, non-obvious, and transferable across domains.", "date": "2026-04-10", "categories": []}
{"id": "2604.10823", "title": "Uncertainty-Guided Attention and Entropy-Weighted Loss for Precise Plant Seedling Segmentation", "abstract": "Plant seedling segmentation supports automated phenotyping in precision agriculture. Standard segmentation models face difficulties due to intricate background images and fine structures in leaves. We introduce UGDA-Net (Uncertainty-Guided Dual Attention Network with Entropy-Weighted Loss and Deep Supervision). Three novel components make up UGDA-Net. The first component is Uncertainty-Guided Dual Attention (UGDA). UGDA uses channel variance to modulate feature maps. The second component is an entropy-weighted hybrid loss function. This loss function focuses on high-uncertainty boundary pixels. The third component employs deep supervision for intermediate encoder layers. We performed a comprehensive systematic ablation study. This study focuses on two widely-used architectures, U-Net and LinkNet. It analyzes five incremental configurations: Baseline, Loss-only, Attention-only, Deep Supervision, and UGDA-Net. We trained UGDA-net using a high-resolution plant seedling image dataset containing 432 images. We demonstrate improved segmentation performance and accuracy. With an increase in Dice coefficient of 9.3% above baseline. LinkNet's variance is 13.2% above baseline. Overlays that are qualitative in nature show the reduced false positives at the leaf boundary. Uncertainty heatmaps are consistent with the complex morphology. UGDA-Net aids in the segmentation of delicate structures in plants and provides a high-def solution. The results showed that uncertainty-guided attention and uncertainty-weighted loss are two complementing systems.", "date": "2026-04-12", "categories": []}
{"id": "2508.07514", "title": "Mitigating Domain Drift in Multi Species Segmentation with DINOv2: A Cross-Domain Evaluation in Herbicide Research Trials", "abstract": "Reliable plant species and damage segmentation for herbicide field research trials requires models that can withstand substantial real-world variation across seasons, geographies, devices, and sensing modalities. Most deep learning approaches trained on controlled datasets fail to generalize under these domain shifts, limiting their suitability for operational phenotyping pipelines. This study evaluates a segmentation framework that integrates vision foundation models (DINOv2) with hierarchical taxonomic inference to improve robustness across heterogeneous agricultural conditions. We train on a large, multi-year dataset collected in Germany and Spain (2018-2020), comprising 14 plant species and 4 herbicide damage classes, and assess generalization under increasingly challenging shifts: temporal and device changes (2023), geographic transfer to the United States, and extreme sensor shift to drone imagery (2024). Results show that the foundation-model backbone consistently outperforms prior baselines, improving species-level F1 from 0.52 to 0.87 on in-distribution data and maintaining significant advantages under moderate (0.77 vs. 0.24) and extreme (0.44 vs. 0.14) shift conditions. Hierarchical inference provides an additional layer of robustness, enabling meaningful predictions even when fine-grained species classification degrades (family F1: 0.68, class F1: 0.88 on aerial imagery). Error analysis reveals that failures under severe shift stem primarily from vegetation-soil confusion, suggesting that taxonomic distinctions remain preserved despite background and viewpoint variability. The system is now deployed within BASF's phenotyping workflow for herbicide research trials across multiple regions, illustrating the practical viability of combining foundation models with structured biological hierarchies for scalable, shift-resilient agricultural monitoring.", "date": "2025-08-11", "categories": []}
{"id": "2604.07814", "title": "AgriChain Visually Grounded Expert Verified Reasoning for Interpretable Agricultural Vision Language Models", "abstract": "Accurate and interpretable plant disease diagnosis remains a major challenge for vision-language models (VLMs) in real-world agriculture. We introduce AgriChain, a dataset of approximately 11,000 expert-curated leaf images spanning diverse crops and pathologies, each paired with (i) a disease label, (ii) a calibrated confidence score (High/Medium/Low), and (iii) an expert-verified chain-of-thought (CoT) rationale. Draft explanations were first generated by GPT-4o and then verified by a professional agricultural engineer using standardized descriptors (e.g., lesion color, margin, and distribution). We fine-tune Qwen2.5-VL-3B on AgriChain, resulting in a specialized model termed AgriChain-VL3B, to jointly predict diseases and generate visually grounded reasoning. On a 1,000-image test set, our CoT-supervised model achieves 73.1% top-1 accuracy (macro F1 = 0.466; weighted F1 = 0.655), outperforming strong baselines including Gemini 1.5 Flash, Gemini 2.5 Pro, and GPT-4o Mini. The generated explanations align closely with expert reasoning, consistently referencing key visual cues. These findings demonstrate that expert-verified reasoning supervision significantly enhances both accuracy and interpretability, bridging the gap between generic multimodal models and human expertise, and advancing trustworthy, globally deployable AI for sustainable agriculture. The dataset and code are publicly available at: https://github.com/hazzanabeel12-netizen/agrichain", "date": "2026-04-09", "categories": []}
{"id": "2604.07182", "title": "TeaLeafVision: An Explainable and Robust Deep Learning Framework for Tea Leaf Disease Classification", "abstract": "As the worlds second most consumed beverage after water, tea is not just a cultural staple but a global economic force of profound scale and influence. More than a mere drink, it represents a quiet negotiation between nature, culture, and the human desire for a moment of reflection. So, the precise identification and detection of tea leaf disease is crucial. With this goal, we have evaluated several Convolutional Neural Networks (CNN) models, among them three shows noticeable performance including DenseNet201, MobileNetV2, InceptionV3 on the teaLeafBD dataset. teaLeafBD dataset contains seven classes, six disease classes and one healthy class, collected under various field conditions reflecting real world challenges. Among the CNN models, DenseNet201 has achieved the highest test accuracy of 99%. In order to enhance the model reliability and interpretability, we have implemented Gradient weighted Class Activation Mapping (Grad CAM), occlusion sensitivity analysis and adversarial training techniques to increase the noise resistance of the model. Finally, we have developed a prototype in order to leverage the models capabilities on real life agriculture. This paper illustrates the deep learning models capabilities to classify the disease in real life tea leaf disease detection and management.", "date": "2026-04-08", "categories": []}
{"id": "2604.05415", "title": "Learning to Synergize Semantic and Geometric Priors for Limited-Data Wheat Disease Segmentation", "abstract": "Wheat disease segmentation is fundamental to precision agriculture but faces severe challenges from significant intra-class temporal variations across growth stages. Such substantial appearance shifts make collecting a representative dataset for training from scratch both labor-intensive and impractical. To address this, we propose SGPer, a Semantic-Geometric Prior Synergization framework that treats wheat disease segmentation under limited data as a coupled task of disease-specific semantic perception and disease boundary localization. Our core insight is that pretrained DINOv2 provides robust category-aware semantic priors to handle appearance shifts, which can be converted into coarse spatial prompts to guide SAM for the precise localization of disease boundaries. Specifically, SGPer designs disease-sensitive adapters with multiple disease-friendly filters and inserts them into both DINOv2 and SAM to align their pretrained representations with disease-specific characteristics. To operationalize this synergy, SGPer transforms DINOv2-derived features into dense, category-specific point prompts to ensure comprehensive spatial coverage of all disease regions. To subsequently eliminate prompt redundancy and ensure highly accurate mask generation, it dynamically filters these dense candidates by cross-referencing SAM's iterative mask confidence with the category-specific semantic consistency derived from DINOv2. Ultimately, SGPer distills a highly informative set of prompts to activate SAM's geometric priors, achieving precise and robust segmentation that remains strictly invariant to temporal appearance changes. Extensive evaluations demonstrate that SGPer consistently achieves state-of-the-art performance on wheat disease and organ segmentation benchmarks, especially in data-constrained scenarios.", "date": "2026-04-07", "categories": []}
{"id": "2507.12590", "title": "From Time-series Generation, Model Selection to Transfer Learning: A Comparative Review of Pixel-wise Approaches for Large-scale Crop Mapping", "abstract": "Crop mapping involves identifying and classifying crop types using spatial data, primarily derived from remote sensing imagery. This study presents the first comprehensive review of large-scale, pixel-wise crop mapping workflows, encompassing both conventional supervised methods and emerging transfer learning approaches. To identify the optimal time-series generation approaches and supervised crop mapping models, we conducted systematic experiments, comparing six widely adopted satellite image-based preprocessing methods, alongside eleven supervised pixel-wise classification models. Additionally, we assessed the synergistic impact of varied training sample sizes and variable combinations. Moreover, we identified optimal transfer learning techniques for different magnitudes of domain shift. The evaluation of optimal methods was conducted across five diverse agricultural sites. Landsat 8 served as the primary satellite data source. Labels come from CDL trusted pixels and field surveys.\n  Our findings reveal three key insights. First, fine-scale interval preprocessing paired with Transformer models consistently delivered optimal performance for both supervised and transferable workflows. RF offered rapid training and competitive performance in conventional supervised learning and direct transfer to similar domains. Second, transfer learning techniques enhanced workflow adaptability, with UDA being effective for homogeneous crop classes while fine-tuning remains robust across diverse scenarios. Finally, workflow choice depends heavily on the availability of labeled samples. With a sufficient sample size, supervised training typically delivers more accurate and generalizable results. Below a certain threshold, transfer learning that matches the level of domain shift is a viable alternative to achieve crop mapping. All code is publicly available to encourage reproducibility practice.", "date": "2025-07-16", "categories": []}
{"id": "2604.03328", "title": "Review and Evaluation of Point-Cloud based Leaf Surface Reconstruction Methods for Agricultural Applications", "abstract": "Accurate reconstruction of leaf surfaces from 3D point cloud is essential for agricultural applications such as phenotyping. However, real-world plant data (i.e., irregular 3D point cloud) are often complex to reconstruct plant parts accurately. A wide range of surface reconstruction methods has been proposed, including parametric, triangulation-based, implicit, and learning based approaches, yet their relative performance for leaf surface reconstruction remains insufficiently understood. In this work, we present a comparative study of nine representative surface reconstruction methods for leaf surfaces. We evaluate these methods on three publicly available datasets: LAST-STRAW, Pheno4D, and Crops3D - spanning diverse species, sensors, and sensing environments, ranging from clean high-resolution indoor scans to noisy low-resolution field settings. The analysis highlights the trade-offs between surface area estimation accuracy, smoothness, robustness to noise and missing data, and computational cost across different methods. These factors affect the cost and constraints of robotic hardware used in agricultural applications. Our results show that each method exhibits distinct advantages depending on application and resource constraints. The findings provide practical guidance for selecting surface reconstruction techniques for resource constrained robotic platforms.", "date": "2026-04-02", "categories": []}
{"id": "2604.00559", "title": "FecalFed: Privacy-Preserving Poultry Disease Detection via Federated Learning", "abstract": "Early detection of highly pathogenic avian influenza (HPAI) and endemic poultry diseases is critical for global food security. While computer vision models excel at classifying diseases from fecal imaging, deploying these systems at scale is bottlenecked by farm data privacy concerns and institutional data silos. Furthermore, existing open-source agricultural datasets frequently suffer from severe, undocumented data contamination. In this paper, we introduce $\\textbf{FecalFed}$, a privacy-preserving federated learning framework for poultry disease classification. We first curate and release $\\texttt{poultry-fecal-fl}$, a rigorously deduplicated dataset of 8,770 unique images across four disease classes, revealing and eliminating a 46.89$\\%$ duplication rate in popular public repositories. To simulate realistic agricultural environments, we evaluate FecalFed under highly heterogeneous, non-IID conditions (Dirichlet $\\alpha=0.5$). While isolated single-farm training collapses under this data heterogeneity, yielding only 64.86$\\%$ accuracy, our federated approach recovers performance without centralizing sensitive data. Specifically, utilizing server-side adaptive optimization (FedAdam) with a Swin-Small architecture achieves 90.31$\\%$ accuracy, closely approaching the centralized upper bound of 95.10\\%. Furthermore, we demonstrate that an edge-optimized Swin-Tiny model maintains highly competitive performance at 89.74$\\%$, establishing a highly efficient, privacy-first blueprint for on-farm avian disease monitoring.", "date": "2026-04-01", "categories": []}
{"id": "2603.26724", "title": "An Annotation-to-Detection Framework for Autonomous and Robust Vine Trunk Localization in the Field by Mobile Agricultural Robots", "abstract": "The dynamic and heterogeneous nature of agricultural fields presents significant challenges for object detection and localization, particularly for autonomous mobile robots that are tasked with surveying previously unseen unstructured environments. Concurrently, there is a growing need for real-time detection systems that do not depend on large-scale manually labeled real-world datasets. In this work, we introduce a comprehensive annotation-to-detection framework designed to train a robust multi-modal detector using limited and partially labeled training data. The proposed methodology incorporates cross-modal annotation transfer and an early-stage sensor fusion pipeline, which, in conjunction with a multi-stage detection architecture, effectively trains and enhances the system's multi-modal detection capabilities. The effectiveness of the framework was demonstrated through vine trunk detection in novel vineyard settings that featured diverse lighting conditions and varying crop densities to validate performance. When integrated with a customized multi-modal LiDAR and Odometry Mapping (LOAM) algorithm and a tree association module, the system demonstrated high-performance trunk localization, successfully identifying over 70% of trees in a single traversal with a mean distance error of less than 0.37m. The results reveal that by leveraging multi-modal, incremental-stage annotation and training, the proposed framework achieves robust detection performance regardless of limited starting annotations, showcasing its potential for real-world and near-ground agricultural applications.", "date": "2026-03-20", "categories": []}
{"id": "2603.27519", "title": "SPROUT: A Scalable Diffusion Foundation Model for Agricultural Vision", "abstract": "Vision Foundation Models (VFM) pre-trained on large-scale unlabeled data have achieved remarkable success on general computer vision tasks, yet typically suffer from significant domain gaps when applied to agriculture. In this context, we introduce $SPROUT$ ($S$calable $P$lant $R$epresentation model via $O$pen-field $U$nsupervised $T$raining), a multi-crop, multi-task agricultural foundation model trained via diffusion denoising. SPROUT leverages a VAE-free Pixel-space Diffusion Transformer to learn rich, structure-aware representations through denoising and enabling efficient end-to-end training. We pre-train SPROUT on a curated dataset of 2.6 million high-quality agricultural images spanning diverse crops, growth stages, and environments. Extensive experiments demonstrate that SPROUT consistently outperforms state-of-the-art web-pretrained and agricultural foundation models across a wide range of downstream tasks, while requiring substantially lower pre-training cost. The code and model are available at https://github.com/UTokyo-FieldPhenomics-Lab/SPROUT.", "date": "2026-03-29", "categories": []}
{"id": "2603.28670", "title": "Sim-to-Real Fruit Detection Using Synthetic Data: Quantitative Evaluation and Embedded Deployment with Isaac Sim", "abstract": "This study investigates the effectiveness of synthetic data for sim-to-real transfer in object detection under constrained data conditions and embedded deployment requirements. Synthetic datasets were generated in NVIDIA Isaac Sim and combined with limited real-world fruit images to train YOLO-based detection models under real-only, synthetic-only, and hybrid regimes. Performance was evaluated on two test datasets: an in-domain dataset with conditions matching the training data and a domain shift dataset containing real fruit and different background conditions. Results show that models trained exclusively on real data achieve the highest accuracy, while synthetic-only models exhibit reduced performance due to a domain gap. Hybrid training strategies significantly improve performance compared to synthetic-only approaches and achieve results close to real-only training while reducing the need for manual annotation. Under domain shift conditions, all models show performance degradation, with hybrid models providing improved robustness. The trained models were successfully deployed on a Jetson Orin NX using TensorRT optimization, achieving real-time inference performance. The findings highlight that synthetic data is most effective when used in combination with real data and that deployment constraints must be considered alongside detection accuracy.", "date": "2026-03-30", "categories": []}
{"id": "2411.16196", "title": "Learn from Foundation Model: Fruit Detection Model without Manual Annotation", "abstract": "Recent breakthroughs in large foundation models have enabled the possibility of transferring knowledge pre-trained on vast datasets to domains with limited data availability. Agriculture is one of the domains that lacks sufficient data. This study proposes a framework to train effective, domain-specific, small models from foundation models without manual annotation. Our approach begins with SDM (Segmentation-Description-Matching), a stage that leverages two foundation models: SAM2 (Segment Anything in Images and Videos) for segmentation and OpenCLIP (Open Contrastive Language-Image Pretraining) for zero-shot open-vocabulary classification. In the second stage, a novel knowledge distillation mechanism is utilized to distill compact, edge-deployable models from SDM, enhancing both inference speed and perception accuracy. The complete method, termed SDM-D (Segmentation-Description-Matching-Distilling), demonstrates strong performance across various fruit detection tasks object detection, semantic segmentation, and instance segmentation) without manual annotation. It nearly matches the performance of models trained with abundant labels. Notably, SDM-D outperforms open-set detection methods such as Grounding SAM and YOLO-World on all tested fruit detection datasets. Additionally, we introduce MegaFruits, a comprehensive fruit segmentation dataset encompassing over 25,000 images, and all code and datasets are made publicly available at https://github.com/AgRoboticsResearch/SDM-D.git.", "date": "2024-11-25", "categories": []}
{"id": "2603.11717", "title": "COTONET: A custom cotton detection algorithm based on YOLO11 for stage of growth cotton boll detection", "abstract": "Cotton harvesting is a critical phase where cotton capsules are physically manipulated and can lead to fibre degradation. To maintain the highest quality, harvesting methods must emulate delicate manual grasping, to preserve cotton's intrinsic properties. Automating this process requires systems capable of recognising cotton capsules across various phenological stages. To address this challenge, we propose COTONET, an enhanced custom YOLO11 model tailored with attention mechanisms to improve the detection of difficult instances. The architecture incorporates gradients in non-learnable operations to enhance shape and feature extraction. Key architectural modifications include: the replacement of convolutional blocks with Squeeze-and-Exitation blocks, a redesigned backbone integrating attention mechanisms, and the substitution of standard upsampling operations for Content Aware Reassembly of Features (CARAFE). Additionally, we integrate Simple Attention Modules (SimAM) for primary feature aggregation and Parallel Hybrid Attention Mechanisms (PHAM) for channel-wise, spatial-wise and coordinate-wise attention in the downward neck path. This configuration offers increased flexibility and robustness for interpreting the complexity of cotton crop growth. COTONET aligns with small-to-medium YOLO models utilizing 7.6M parameters and 27.8 GFLOPS, making it suitable for low-resource edge computing and mobile robotics. COTONET outperforms the standard YOLO baselines, achieving a mAP50 of 81.1% and a mAP50-95 of 60.6%.", "date": "2026-03-12", "categories": []}
{"id": "2603.01932", "title": "BAWSeg: A UAV Multispectral Benchmark for Barley Weed Segmentation", "abstract": "Accurate weed mapping in cereal fields requires pixel-level segmentation from UAV imagery that remains reliable across fields, seasons, and illumination. Existing multispectral pipelines often depend on thresholded vegetation indices, which are brittle under radiometric drift and mixed crop--weed pixels, or on single-stream CNN and Transformer backbones that ingest stacked bands and indices, where radiance cues and normalized index cues interfere and reduce sensitivity to small weed clusters embedded in crop canopy. We propose VISA, a two-stream segmentation network that decouples these cues and fuses them at native resolution. The radiance stream learns from calibrated five-band reflectance using local residual convolutions, channel recalibration, spatial gating, and skip-connected decoding, which preserve fine textures, row boundaries, and small weed structures that are often weakened after ratio-based index compression. The index stream operates on vegetation-index maps with windowed self-attention to model local structure efficiently, state-space layers to propagate field-scale context without quadratic attention cost, and Slot Attention to form stable region descriptors that improve discrimination of sparse weeds under canopy mixing. To support supervised training and deployment-oriented evaluation, we introduce BAWSeg, a four-year UAV multispectral dataset collected over commercial barley paddocks in Western Australia, providing radiometrically calibrated blue, green, red, red edge, and near-infrared orthomosaics, derived vegetation indices, and dense crop, weed, and other labels with leakage-free block splits. On BAWSeg, VISA achieves 75.6% mIoU and 63.5% weed IoU with 22.8 M parameters, outperforming a multispectral SegFormer-B1 baseline by 1.2 mIoU and 1.9 weed IoU. Under cross-plot and cross-year protocols, VISA maintains 71.2% and 69.2% mIoU, respectively.", "date": "2026-03-02", "categories": []}
{"id": "2603.07356", "title": "AgrI Challenge: A Data-Centric AI Competition for Cross-Team Validation in Agricultural Vision", "abstract": "Machine learning models in agricultural vision often achieve high accuracy on curated datasets but fail to generalize under real field conditions due to distribution shifts between training and deployment environments. Moreover, most machine learning competitions focus primarily on model design while treating datasets as fixed resources, leaving the role of data collection practices in model generalization largely unexplored.\n  We introduce the AgrI Challenge, a data-centric competition framework in which multiple teams independently collect field datasets, producing a heterogeneous multi-source benchmark that reflects realistic variability in acquisition conditions. To systematically evaluate cross-domain generalization across independently collected datasets, we propose Cross-Team Validation (CTV), an evaluation paradigm that treats each team's dataset as a distinct domain. CTV includes two complementary protocols: Train-on-One-Team-Only (TOTO), which measures single-source generalization, and Leave-One-Team-Out (LOTO), which evaluates collaborative multi-source training.\n  Experiments reveal substantial generalization gaps under single-source training: models achieve near-perfect validation accuracy yet exhibit validation-test gaps of up to 16.20% (DenseNet121) and 11.37% (Swin Transformer) when evaluated on datasets collected by other teams. In contrast, collaborative multi-source training dramatically improves robustness, reducing the gap to 2.82% and 1.78%, respectively.\n  The challenge also produced a publicly available dataset of 50,673 field images of six tree species collected by twelve independent teams, providing a diverse benchmark for studying domain shift and data-centric learning in agricultural vision.", "date": "2026-03-07", "categories": []}
{"id": "2510.21732", "title": "Automated Pest Counting in Water Traps through Active Robotic Stirring for Occlusion Handling", "abstract": "Existing image-based pest counting methods rely on single static images and often produce inaccurate results under occlusion. To address this issue, this paper proposes an automated pest counting method in water traps through active robotic stirring. First, an automated robotic arm-based stirring system is developed to redistribute pests and reveal occluded individuals for counting. Then, the effects of different stirring patterns on pest counting performance are investigated. Six stirring patterns are designed and evaluated across different pest density scenarios to identify the optimal one. Finally, a heuristic counting confidence-driven closed-loop control system is proposed for adaptive-speed robotic stirring, adjusting the stirring speed based on the average change rate of counting confidence between consecutive frames. Experimental results show that the four circles is the optimal stirring pattern, achieving the lowest overall mean absolute counting error of 4.384 and the highest overall mean counting confidence of 0.721. Compared with constant-speed stirring, adaptive-speed stirring reduces task execution time by up to 44.7% and achieves more stable performance across different pest density scenarios. Moreover, the proposed pest counting method reduces the mean absolute counting error by up to 3.428 compared to the single static image counting method under high-density scenarios where occlusion is severe.", "date": "2025-09-30", "categories": []}
{"id": "2601.05143", "title": "A Two-Stage Multitask Vision-Language Framework for Explainable Crop Disease Visual Question Answering", "abstract": "Visual question answering (VQA) for crop disease analysis requires accurate visual understanding and reliable language generation. In this work, we present a lightweight and explainable vision-language framework for crop and disease identification from leaf images. The proposed approach integrates a Swin Transformer vision encoder with sequence-to-sequence language decoders. The vision encoder is first trained in a multitask setup for both plant and disease classification, and then frozen while the text decoders are trained, forming a two-stage training strategy that enhances visual representation learning and cross-modal alignment. We evaluate the model on the large-scale Crop Disease Domain Multimodal (CDDM) dataset using both classification and natural language generation metrics. Experimental results demonstrate near-perfect recognition performance, achieving 99.94% plant classification accuracy and 99.06% disease classification accuracy, along with strong BLEU, ROUGE and BERTScore results. Without fine-tuning, the model further generalizes well to the external PlantVillageVQA benchmark, achieving 83.18% micro accuracy in the VQA task. Our lightweight design outperforms larger vision-language baselines while using significantly fewer parameters. Explainability is assessed through Grad-CAM and token-level attribution, providing interpretable visual and textual evidence for predictions. Qualitative results demonstrate robust performance under diverse user-driven queries, highlighting the effectiveness of task-specific visual pretraining and the two-stage training methodology for crop disease visual question answering. An interactive demo of the proposed Swin-T5 model is publicly available as a Gradio-based application at https://huggingface.co/spaces/Zahid16/PlantDiseaseVQAwithSwinT5 for community use.", "date": "2026-01-08", "categories": []}
{"id": "2603.02419", "title": "DINOv3 Visual Representations for Blueberry Perception Toward Robotic Harvesting", "abstract": "Vision Foundation Models trained via large-scale self-supervised learning have demonstrated strong generalization in visual perception; however, their practical role and performance limits in agricultural settings remain insufficiently understood. This work evaluates DINOv3 as a frozen backbone for blueberry robotic harvesting-related visual tasks, including fruit and bruise segmentation, as well as fruit and cluster detection. Under a unified protocol with lightweight decoders, segmentation benefits consistently from stable patch-level representations and scales with backbone size. In contrast, detection is constrained by target scale variation, patch discretization, and localization compatibility. The failure of cluster detection highlights limitations in modeling relational targets defined by spatial aggregation. Overall, DINOv3 is best viewed not as an end-to-end task model, but as a semantic backbone whose effectiveness depends on downstream spatial modeling aligned with fruit-scale and aggregation structures, providing guidance for blueberry robotic harvesting. Code and dataset will be available upon acceptance.", "date": "2026-03-02", "categories": []}
{"id": "2603.06655", "title": "A Parameter-efficient Convolutional Approach for Weed Detection in Multispectral Aerial Imagery", "abstract": "We introduce FCBNet, an efficient model designed for weed segmentation. The architecture is based on a fully frozen ConvNeXt backbone, the proposed Feature Correction Block (FCB), which leverages efficient convolutions for feature refinement, and a lightweight decoder. FCBNet is evaluated on the WeedBananaCOD and WeedMap datasets under both RGB and multispectral modalities, showing that FCBNet outperforms models such as U-Net, DeepLabV3+, SK-U-Net, SegFormer, and WeedSense in terms of mIoU, exceeding 85%, while also achieving superior computational efficiency, requiring only 0.06 to 0.2 hours for training. Furthermore, the frozen backbone strategy reduces the number of trainable parameters by more than 90%, significantly lowering memory requirements.", "date": "2026-02-28", "categories": []}
{"id": "2603.06512", "title": "SG-DOR: Learning Scene Graphs with Direction-Conditioned Occlusion Reasoning for Pepper Plants", "abstract": "Robotic harvesting in dense crop canopies requires effective interventions that depend not only on geometry, but also on explicit, direction-conditioned relations identifying which organs obstruct a target fruit. We present SG-DOR (Scene Graphs with Direction-Conditioned Occlusion Reasoning), a relational framework that, given instance-segmented organ point clouds, infers a scene graph encoding physical attachments and direction-conditioned occlusion. We introduce an occlusion ranking task for retrieving and ranking candidate leaves for a target fruit and approach direction, and propose a direction-aware graph neural architecture with per-fruit leaf-set attention and union-level aggregation. Experiments on a multi-plant synthetic pepper dataset show improved occlusion prediction (F1=0.73, NDCG@3=0.85) and attachment inference (edge F1=0.83) over strong ablations, yielding a structured relational signal for downstream intervention planning.", "date": "2026-03-06", "categories": []}
{"id": "2603.04091", "title": "CLIP-Guided Multi-Task Regression for Multi-View Plant Phenotyping", "abstract": "Modeling plant growth dynamics plays a central role in modern agricultural research. However, learning robust predictors from multi-view plant imagery remains challenging due to strong viewpoint redundancy and viewpoint-dependent appearance changes. We propose a level-aware vision language framework that jointly predicts plant age and leaf count using a single multi-task model built on CLIP embeddings. Our method aggregates rotational views into angle-invariant representations and conditions visual features on lightweight text priors encoding viewpoint level for stable prediction under incomplete or unordered inputs. On the GroMo25 benchmark, our approach reduces mean age MAE from 7.74 to 3.91 and mean leaf-count MAE from 5.52 to 3.08 compared to the GroMo baseline, corresponding to improvements of 49.5% and 44.2%, respectively. The unified formulation simplifies the pipeline by replacing the conventional dual-model setup while improving robustness to missing views. The models and code is available at: https://github.com/SimonWarmers/CLIP-MVP", "date": "2026-03-04", "categories": []}
{"id": "2603.00160", "title": "DINOv3 Meets YOLO26 for Weed Detection in Vegetable Crops", "abstract": "Developing robust models for precision vegetable weeding is currently constrained by the scarcity of large-scale, annotated weed-crop datasets. To address this limitation, this study proposes a foundational crop-weed detection model by integrating heterogeneous datasets and leveraging self-supervised learning. A total of 618,642 crop-weed images were initially collected and subsequently refined to 199,388 filtered images for fine-tuning a DINOv3 vision transformer (ViT-small) through a sequential curation strategy. The fine-tuned DINOv3 backbone was then integrated into YOLO26, serving either as a primary backbone or part of a dual-backbone architecture. A feature alignment loss was introduced in the dual backbone framework to enhance feature fusion with minimal computational overhead. Experimental results show that the proposed DINOv3-finetuned ViT-small-based YOLO26-large achieved up to a +5.4% mAP50 gain on in-domain images collected in the 2025 season. Moreover, it demonstrated strong cross-domain generalization with mAP50 improvements of +14.0% on the 2021-2023 season dataset and +11.9% on the 2024 season dataset, compared to the standard YOLO26-large. Although the DINOv3-YOLO26-large model has 45.6% more parameters and a 2.9x increase in inference latency, it maintains real-time performance at ~28.5 frames per second (fps). The curated dataset and software programs developed in this study will be made publicly available.", "date": "2026-02-25", "categories": []}
{"id": "2603.00168", "title": "Image-Based Classification of Olive Species Specific to Turkiye with Deep Neural Networks", "abstract": "In this study, image processing and deep learning methodologies were employed to automatically classify local olive species cultivated in Turkiye. A stereo camera was utilized to capture images of five distinct olive species, which were then preprocessed to ensure their suitability for analysis. Convolutional Neural Network (CNN) architectures, specifically MobileNetV2 and EfficientNetB0, were employed for image classification. These models were optimized through a transfer learning approach. The training and testing results indicated that the EfficientNetB0 model exhibited the optimal performance, with an accuracy of 94.5%. The findings demonstrate that deep learning-based systems offer an effective solution for classifying olive species with high accuracy. The developed method has significant potential for application in areas such as automatic identification and quality control of agricultural products.", "date": "2026-02-26", "categories": []}
{"id": "2602.17372", "title": "Tree crop mapping of South America reveals links to deforestation and conservation", "abstract": "Monitoring tree crop expansion is vital for zero-deforestation policies like the European Union's Regulation on Deforestation-free Products (EUDR). However, these efforts are hindered by a lack of highresolution data distinguishing diverse agricultural systems from forests. Here, we present the first 10m-resolution tree crop map for South America, generated using a multi-modal, spatio-temporal deep learning model trained on Sentinel-1 and Sentinel-2 satellite imagery time series. The map identifies approximately 11 million hectares of tree crops, 23% of which is linked to 2000-2020 forest cover loss. Critically, our analysis reveals that existing regulatory maps supporting the EUDR often classify established agriculture, particularly smallholder agroforestry, as \"forest\". This discrepancy risks false deforestation alerts and unfair penalties for small-scale farmers. Our work mitigates this risk by providing a high-resolution baseline, supporting conservation policies that are effective, inclusive, and equitable.", "date": "2026-02-19", "categories": []}
{"id": "2602.15656", "title": "A Novel Public Dataset for Strawberry (Fragaria x ananassa) Ripeness Detection and Comparative Evaluation of YOLO-Based Models", "abstract": "The strawberry (Fragaria x ananassa), known worldwide for its economic value and nutritional richness, is a widely cultivated fruit. Determining the correct ripeness level during the harvest period is crucial for both preventing losses for producers and ensuring consumers receive a quality product. However, traditional methods, i.e., visual assessments alone, can be subjective and have a high margin of error. Therefore, computer-assisted systems are needed. However, the scarcity of comprehensive datasets accessible to everyone in the literature makes it difficult to compare studies in this field. In this study, a new and publicly available strawberry ripeness dataset, consisting of 566 images and 1,201 labeled objects, prepared under variable light and environmental conditions in two different greenhouses in Turkey, is presented to the literature. Comparative tests conducted on the data set using YOLOv8, YOLOv9, and YOLO11-based models showed that the highest precision value was 90.94% in the YOLOv9c model, while the highest recall value was 83.74% in the YOLO11s model. In terms of the general performance criterion mAP@50, YOLOv8s was the best performing model with a success rate of 86.09%. The results show that small and medium-sized models work more balanced and efficiently on this type of dataset, while also establishing a fundamental reference point for smart agriculture applications.", "date": "2026-02-17", "categories": []}
{"id": "2602.18530", "title": "Image-Based Classification of Olive Varieties Native to Turkiye Using Multiple Deep Learning Architectures: Analysis of Performance, Complexity, and Generalization", "abstract": "This study compares multiple deep learning architectures for the automated, image-based classification of five locally cultivated black table olive varieties in Turkey: Gemlik, Ayvalik, Uslu, Erkence, and Celebi. Using a dataset of 2500 images, ten architectures - MobileNetV2, EfficientNetB0, EfficientNetV2-S, ResNet50, ResNet101, DenseNet121, InceptionV3, ConvNeXt-Tiny, ViT-B16, and Swin-T - were trained using transfer learning. Model performance was evaluated using accuracy, precision, recall, F1-score, Matthews Correlation Coefficient (MCC), Cohen's Kappa, ROC-AUC, number of parameters, FLOPs, inference time, and generalization gap. EfficientNetV2-S achieved the highest classification accuracy (95.8%), while EfficientNetB0 provided the best trade-off between accuracy and computational complexity. Overall, the results indicate that under limited data conditions, parametric efficiency plays a more critical role than model depth alone.", "date": "2026-02-20", "categories": []}
{"id": "2602.16749", "title": "U-FedTomAtt: Ultra-lightweight Federated Learning with Attention for Tomato Disease Recognition", "abstract": "Federated learning has emerged as a privacy-preserving and efficient approach for deploying intelligent agricultural solutions. Accurate edge-based diagnosis across geographically dispersed farms is crucial for recognising tomato diseases in sustainable farming. Traditional centralised training aggregates raw data on a central server, leading to communication overhead, privacy risks and latency. Meanwhile, edge devices require lightweight networks to operate effectively within limited resources. In this paper, we propose U-FedTomAtt, an ultra-lightweight federated learning framework with attention for tomato disease recognition in resource-constrained and distributed environments. The model comprises only 245.34K parameters and 71.41 MFLOPS. First, we propose an ultra-lightweight neural network with dilated bottleneck (DBNeck) modules and a linear transformer to minimise computational and memory overhead. To mitigate potential accuracy loss, a novel local-global residual attention (LoGRA) module is incorporated. Second, we propose the federated dual adaptive weight aggregation (FedDAWA) algorithm that enhances global model accuracy. Third, our framework is validated using three benchmark datasets for tomato diseases under simulated federated settings. Experimental results show that the proposed method achieves 0.9910% and 0.9915% Top-1 accuracy and 0.9923% and 0.9897% F1-scores on SLIF-Tomato and PlantVillage tomato datasets, respectively.", "date": "2026-02-18", "categories": []}
{"id": "2602.14140", "title": "Detection of On-Ground Chestnuts Using Artificial Intelligence Toward Automated Picking", "abstract": "Traditional mechanized chestnut harvesting is too costly for small producers, non-selective, and prone to damaging nuts. Accurate, reliable detection of chestnuts on the orchard floor is crucial for developing low-cost, vision-guided automated harvesting technology. However, developing a reliable chestnut detection system faces challenges in complex environments with shading, varying natural light conditions, and interference from weeds, fallen leaves, stones, and other foreign on-ground objects, which have remained unaddressed. This study collected 319 images of chestnuts on the orchard floor, containing 6524 annotated chestnuts. A comprehensive set of 29 state-of-the-art real-time object detectors, including 14 in the YOLO (v11-13) and 15 in the RT-DETR (v1-v4) families at varied model scales, was systematically evaluated through replicated modeling experiments for chestnut detection. Experimental results show that the YOLOv12m model achieves the best mAP@0.5 of 95.1% among all the evaluated models, while the RT-DETRv2-R101 was the most accurate variant among RT-DETR models, with mAP@0.5 of 91.1%. In terms of mAP@[0.5:0.95], the YOLOv11x model achieved the best accuracy of 80.1%. All models demonstrate significant potential for real-time chestnut detection, and YOLO models outperformed RT-DETR models in terms of both detection accuracy and inference, making them better suited for on-board deployment. Both the dataset and software programs in this study have been made publicly available at https://github.com/AgFood-Sensing-and-Intelligence-Lab/ChestnutDetection.", "date": "2026-02-15", "categories": []}
{"id": "2602.10967", "title": "Healthy Harvests: A Comparative Look at Guava Disease Classification Using InceptionV3", "abstract": "Guava fruits often suffer from many diseases. This can harm fruit quality and fruit crop yield. Early identification is important for minimizing damage and ensuring fruit health. This study focuses on 3 different categories for classifying diseases. These are Anthracnose, Fruit flies, and Healthy fruit. The data set used in this study is collected from Mendeley Data. This dataset contains 473 original images of Guava. These images vary in size and format. The original dataset was resized to 256x256 pixels with RGB color mode for better consistency. After this, the Data augmentation process is applied to improve the dataset by generating variations of the original images. The augmented dataset consists of 3784 images using advanced preprocessing techniques. Two deep learning models were implemented to classify the images. The InceptionV3 model is well known for its advanced framework. These apply multiple convolutional filters for obtaining different features effectively. On the other hand, the ResNet50 model helps to train deeper networks by using residual learning. The InceptionV3 model achieved the impressive accuracy of 98.15%, and ResNet50got 94.46% accuracy. Data mixing methods such as CutMix and MixUp were applied to enhance the model's robustness. The confusion matrix was used to evaluate the overall model performance of both InceptionV3 and Resnet50. Additionally, SHAP analysis is used to improve interpretability, which helps to find the significant parts of the image for the model prediction. This study purposes to highlight how advanced models enhan", "date": "2026-02-11", "categories": []}
{"id": "2503.03200", "title": "Transformer-Based Spatio-Temporal Association of Apple Fruitlets", "abstract": "In this paper, we present a transformer-based method to spatio-temporally associate apple fruitlets in stereo-images collected on different days and from different camera poses. State-of-the-art association methods in agriculture are dedicated towards matching larger crops using either high-resolution point clouds or temporally stable features, which are both difficult to obtain for smaller fruit in the field. To address these challenges, we propose a transformer-based architecture that encodes the shape and position of each fruitlet, and propagates and refines these features through a series of transformer encoder layers with alternating self and cross-attention. We demonstrate that our method is able to achieve an F1-score of 92.4% on data collected in a commercial apple orchard and outperforms all baselines and ablations.", "date": "2025-03-05", "categories": []}
{"id": "2407.12040", "title": "Comprehensive Performance Evaluation of YOLOv12, YOLO11, YOLOv10, YOLOv9 and YOLOv8 on Detecting and Counting Fruitlet in Complex Orchard Environments", "abstract": "This study systematically conducted an extensive real-world evaluation of all configurations of You Only Look Once (YOLO)-based object detection algorithms, including YOLOv8, YOLOv9, YOLOv10, YOLO11, and YOLOv12. Models were assessed using precision, recall, mean Average Precision at 50 % Intersection over Union (mAP@50), and computational efficiency across pre-processing, inference, and post-processing stages for detecting immature green fruitlets in commercial orchards. Field-level fruitlet counting was also validated using images captured with both Intel RealSense and iPhone 14 Pro Max sensors. YOLOv12l achieved the highest recall (0.900), while YOLOv10x and YOLOv9 GELAN-c reported the top precision scores of 0.908 and 0.903, respectively. YOLOv9 GELAN-base and GELAN-e achieved the highest mAP@50 (0.935), followed by YOLO11s (0.933) and YOLOv12l (0.931). In counting validation, YOLO11n demonstrated superior accuracy, with RMSE values of 4.51-4.96 and MAE values of 3.85-7.73 across four apple varieties. Sensor-specific training on Intel RealSense further improved detection performance. YOLO11n also recorded the fastest inference speed (2.4 ms), outperforming YOLOv8n, YOLOv9 GELAN-s, YOLOv10n, and YOLOv12n, affirming its suitability for real-time orchard applications.", "date": "2024-07-01", "categories": []}
{"id": "2510.09948", "title": "A Multi-Strategy Framework for Enhancing Shatian Pomelo Detection in Real-World Orchards", "abstract": "Shatian pomelo detection in orchards is essential for yield estimation and lean production, but models tuned to ideal datasets often degrade in practice due to device-dependent tone shifts, illumination changes, large scale variation, and frequent occlusion. We introduce STP-AgriData, a multi-scenario dataset combining real-orchard imagery with curated web images, and apply contrast/brightness augmentations to emulate unstable lighting. To better address scale and occlusion, we propose REAS-Det, featuring Global-Selective Visibility Convolution (GSV-Conv) that expands the visible feature space under global semantic guidance while retaining efficient spatial aggregation, plus C3RFEM, MultiSEAM, and Soft-NMS for refined separation and localization. On STP-AgriData, REAS-Det achieves 86.5% precision, 77.2% recall, 84.3% mAP@0.50, and 53.6% mAP@0.50:0.95, outperforming recent detectors and improving robustness in real orchard environments. The source code is available at: https://github.com/Genk641/REAS-Det.", "date": "2025-10-11", "categories": []}
{"id": "2602.08131", "title": "Fields of The World: A Field Guide for Extracting Agricultural Field Boundaries", "abstract": "Field boundary maps are a building block for agricultural data products and support crop monitoring, yield estimation, and disease estimation. This tutorial presents the Fields of The World (FTW) ecosystem: a benchmark of 1.6M field polygons across 24 countries, pre-trained segmentation models, and command-line inference tools. We provide two notebooks that cover (1) local-scale field boundary extraction with crop classification and forest loss attribution, and (2) country-scale inference using cloud-optimized data. We use MOSAIKS random convolutional features and FTW derived field boundaries to map crop type at the field level and report macro F1 scores of 0.65--0.75 for crop type classification with limited labels. Finally, we show how to explore pre-computed predictions over five countries (4.76M km\\textsuperscript{2}), with median predicted field areas from 0.06 ha (Rwanda) to 0.28 ha (Switzerland).", "date": "2026-02-08", "categories": []}
{"id": "2602.08168", "title": "DAS-SK: An Adaptive Model Integrating Dual Atrous Separable and Selective Kernel CNN for Agriculture Semantic Segmentation", "abstract": "Semantic segmentation in high-resolution agricultural imagery demands models that strike a careful balance between accuracy and computational efficiency to enable deployment in practical systems. In this work, we propose DAS-SK, a novel lightweight architecture that retrofits selective kernel convolution (SK-Conv) into the dual atrous separable convolution (DAS-Conv) module to strengthen multi-scale feature learning. The model further enhances the atrous spatial pyramid pooling (ASPP) module, enabling the capture of fine-grained local structures alongside global contextual information. Built upon a modified DeepLabV3 framework with two complementary backbones - MobileNetV3-Large and EfficientNet-B3, the DAS-SK model mitigates limitations associated with large dataset requirements, limited spectral generalization, and the high computational cost that typically restricts deployment on UAVs and other edge devices. Comprehensive experiments across three benchmarks: LandCover.ai, VDD, and PhenoBench, demonstrate that DAS-SK consistently achieves state-of-the-art performance, while being more efficient than CNN-, transformer-, and hybrid-based competitors. Notably, DAS-SK requires up to 21x fewer parameters and 19x fewer GFLOPs than top-performing transformer models. These findings establish DAS-SK as a robust, efficient, and scalable solution for real-time agricultural robotics and high-resolution remote sensing, with strong potential for broader deployment in other vision domains.", "date": "2026-02-09", "categories": []}
{"id": "2503.11692", "title": "FloPE: Flower Pose Estimation for Precision Pollination", "abstract": "This study presents Flower Pose Estimation (FloPE), a real-time flower pose estimation framework for computationally constrained robotic pollination systems. Robotic pollination has been proposed to supplement natural pollination to ensure global food security due to the decreased population of natural pollinators. However, flower pose estimation for pollination is challenging due to natural variability, flower clusters, and high accuracy demands due to the flowers' fragility when pollinating. This method leverages 3D Gaussian Splatting to generate photorealistic synthetic datasets with precise pose annotations, enabling effective knowledge distillation from a high-capacity teacher model to a lightweight student model for efficient inference. The approach was evaluated on both single and multi-arm robotic platforms, achieving a mean pose estimation error of 0.6 cm and 19.14 degrees within a low computational cost. Our experiments validate the effectiveness of FloPE, achieving up to 78.75% pollination success rate and outperforming prior robotic pollination techniques.", "date": "2025-03-08", "categories": []}
{"id": "2602.06786", "title": "Machine Learning for Detection and Severity Estimation of Sweetpotato Weevil Damage in Field and Lab Conditions", "abstract": "Sweetpotato weevils (Cylas spp.) are considered among the most destructive pests impacting sweetpotato production, particularly in sub-Saharan Africa. Traditional methods for assessing weevil damage, predominantly relying on manual scoring, are labour-intensive, subjective, and often yield inconsistent results. These challenges significantly hinder breeding programs aimed at developing resilient sweetpotato varieties. This study introduces a computer vision-based approach for the automated evaluation of weevil damage in both field and laboratory contexts. In the field settings, we collected data to train classification models to predict root-damage severity levels, achieving a test accuracy of 71.43%. Additionally, we established a laboratory dataset and designed an object detection pipeline employing YOLO12, a leading real-time detection model. This methodology incorporated a two-stage laboratory pipeline that combined root segmentation with a tiling strategy to improve the detectability of small objects. The resulting model demonstrated a mean average precision of 77.7% in identifying minute weevil feeding holes. Our findings indicate that computer vision technologies can provide efficient, objective, and scalable assessment tools that align seamlessly with contemporary breeding workflows. These advancements represent a significant improvement in enhancing phenotyping efficiency within sweetpotato breeding programs and play a crucial role in mitigating the detrimental effects of weevils on food security.", "date": "2026-02-06", "categories": []}
{"id": "2601.00645", "title": "Quality Detection of Stored Potatoes via Transfer Learning: A CNN and Vision Transformer Approach", "abstract": "Image-based deep learning provides a non-invasive, scalable solution for monitoring potato quality during storage, addressing key challenges such as sprout detection, weight loss estimation, and shelf-life prediction. In this study, images and corresponding weight data were collected over a 200-day period under controlled temperature and humidity conditions. Leveraging powerful pre-trained architectures of ResNet, VGG, DenseNet, and Vision Transformer (ViT), we designed two specialized models: (1) a high-precision binary classifier for sprout detection, and (2) an advanced multi-class predictor to estimate weight loss and forecast remaining shelf-life with remarkable accuracy. DenseNet achieved exceptional performance, with 98.03% accuracy in sprout detection. Shelf-life prediction models performed best with coarse class divisions (2-5 classes), achieving over 89.83% accuracy, while accuracy declined for finer divisions (6-8 classes) due to subtle visual differences and limited data per class. These findings demonstrate the feasibility of integrating image-based models into automated sorting and inventory systems, enabling early identification of sprouted potatoes and dynamic categorization based on storage stage. Practical implications include improved inventory management, differential pricing strategies, and reduced food waste across supply chains. While predicting exact shelf-life intervals remains challenging, focusing on broader class divisions ensures robust performance. Future research should aim to develop generalized models trained on diverse potato varieties and storage conditions to enhance adaptability and scalability. Overall, this approach offers a cost-effective, non-destructive method for quality assessment, supporting efficiency and sustainability in potato storage and distribution.", "date": "2026-01-02", "categories": []}
{"id": "2507.01269", "title": "Advancements in Weed Mapping: A Systematic Review", "abstract": "Weed mapping plays a critical role in precision management by providing accurate and timely data on weed distribution, enabling targeted control and reduced herbicide use. This minimizes environmental impacts, supports sustainable land management, and improves outcomes across agricultural and natural environments. Recent advances in weed mapping leverage ground-vehicle Red Green Blue (RGB) cameras, satellite and drone-based remote sensing combined with sensors such as spectral, Near Infra-Red (NIR), and thermal cameras. The resulting data are processed using advanced techniques including big data analytics and machine learning, significantly improving the spatial and temporal resolution of weed maps and enabling site-specific management decisions. Despite a growing body of research in this domain, there is a lack of comprehensive literature reviews specifically focused on weed mapping. In particular, the absence of a structured analysis spanning the entire mapping pipeline, from data acquisition to processing techniques and mapping tools, limits progress in the field. This review addresses these gaps by systematically examining state-of-the-art methods in data acquisition (sensor and platform technologies), data processing (including annotation and modelling), and mapping techniques (such as spatiotemporal analysis and decision support tools). Following PRISMA guidelines, we critically evaluate and synthesize key findings from the literature to provide a holistic understanding of the weed mapping landscape. This review serves as a foundational reference to guide future research and support the development of efficient, scalable, and sustainable weed management systems.", "date": "2025-07-02", "categories": []}
{"id": "2602.00216", "title": "Development of a Cacao Disease Identification and Management App Using Deep Learning", "abstract": "Smallholder cacao producers often rely on outdated farming techniques and face significant challenges from pests and diseases, unlike larger plantations with more resources and expertise. In the Philippines, cacao farmers have limited access to data, information, and good agricultural practices. This study addresses these issues by developing a mobile application for cacao disease identification and management that functions offline, enabling use in remote areas where farms are mostly located. The core of the system is a deep learning model trained to identify cacao diseases accurately. The trained model is integrated into the mobile app to support farmers in field diagnosis. The disease identification model achieved a validation accuracy of 96.93% while the model for detecting cacao black pod infection levels achieved 79.49% validation accuracy. Field testing of the application showed an agreement rate of 84.2% compared with expert cacao technician assessments. This approach empowers smallholder farmers by providing accessible, technology-enabled tools to improve cacao crop health and productivity.", "date": "2026-01-30", "categories": []}
{"id": "2602.00703", "title": "StomataSeg: Semi-Supervised Instance Segmentation for Sorghum Stomatal Components", "abstract": "Sorghum is a globally important cereal grown widely in water-limited and stress-prone regions. Its strong drought tolerance makes it a priority crop for climate-resilient agriculture. Improving water-use efficiency in sorghum requires precise characterisation of stomatal traits, as stomata control of gas exchange, transpiration and photosynthesis have a major influence on crop performance. Automated analysis of sorghum stomata is difficult because the stomata are small (often less than 40 $\\mu$m in length in grasses such as sorghum) and vary in shape across genotypes and leaf surfaces. Automated segmentation contributes to high-throughput stomatal phenotyping, yet current methods still face challenges related to nested small structures and annotation bottlenecks. In this paper, we propose a semi-supervised instance segmentation framework tailored for analysis of sorghum stomatal components. We collect and annotate a sorghum leaf imagery dataset containing 11,060 human-annotated patches, covering the three stomatal components (pore, guard cell and complex area) across multiple genotypes and leaf surfaces. To improve the detection of tiny structures, we split high-resolution microscopy images into overlapping small patches. We then apply a pseudo-labelling strategy to unannotated images, producing an additional 56,428 pseudo-labelled patches. Benchmarking across semantic and instance segmentation models shows substantial performance gains: for semantic models the top mIoU increases from 65.93% to 70.35%, whereas for instance models the top AP rises from 28.30% to 46.10%. These results demonstrate that combining patch-based preprocessing with semi-supervised learning significantly improves the segmentation of fine stomatal structures. The proposed framework supports scalable extraction of stomatal traits and facilitates broader adoption of AI-driven phenotyping in crop science.", "date": "2026-01-31", "categories": []}
{"id": "2601.22732", "title": "Active Learning-Driven Lightweight YOLOv9: Enhancing Efficiency in Smart Agriculture", "abstract": "This study addresses the demand for real-time detection of tomatoes and tomato flowers by agricultural robots deployed on edge devices in greenhouse environments. Under practical imaging conditions, object detection systems often face challenges such as large scale variations caused by varying camera distances, severe occlusion from plant structures, and highly imbalanced class distributions. These factors make conventional object detection approaches that rely on fully annotated datasets difficult to simultaneously achieve high detection accuracy and deployment efficiency. To overcome these limitations, this research proposes an active learning driven lightweight object detection framework, integrating data analysis, model design, and training strategy. First, the size distribution of objects in raw agricultural images is analyzed to redefine an operational target range, thereby improving learning stability under real-world conditions. Second, an efficient feature extraction module is incorporated to reduce computational cost, while a lightweight attention mechanism is introduced to enhance feature representation under multi-scale and occluded scenarios. Finally, an active learning strategy is employed to iteratively select high-information samples for annotation and training under a limited labeling budget, effectively improving the recognition performance of minority and small-object categories. Experimental results demonstrate that, while maintaining a low parameter count and inference cost suitable for edge-device deployment, the proposed method effectively improves the detection performance of tomatoes and tomato flowers in raw images. Under limited annotation conditions, the framework achieves an overall detection accuracy of 67.8% mAP, validating its practicality and feasibility for intelligent agricultural applications.", "date": "2026-01-30", "categories": []}
{"id": "2407.21666", "title": "An explainable vision transformer with transfer learning based efficient drought stress identification", "abstract": "Early detection of drought stress is critical for taking timely measures for reducing crop loss before the drought impact becomes irreversible. The subtle phenotypical and physiological changes in response to drought stress are captured by non-invasive imaging techniques and these imaging data serve as valuable resource for machine learning methods to identify drought stress. While convolutional neural networks (CNNs) are in wide use, vision transformers (ViTs) present a promising alternative in capturing long-range dependencies and intricate spatial relationships, thereby enhancing the detection of subtle indicators of drought stress. We propose an explainable deep learning pipeline that leverages the power of ViTs for drought stress detection in potato crops using aerial imagery. We applied two distinct approaches: a synergistic combination of ViT and support vector machine (SVM), where ViT extracts intricate spatial features from aerial images, and SVM classifies the crops as stressed or healthy and an end-to-end approach using a dedicated classification layer within ViT to directly detect drought stress. Our key findings explain the ViT model's decision-making process by visualizing attention maps. These maps highlight the specific spatial features within the aerial images that the ViT model focuses as the drought stress signature. Our findings demonstrate that the proposed methods not only achieve high accuracy in drought stress identification but also shedding light on the diverse subtle plant features associated with drought stress. This offers a robust and interpretable solution for drought stress monitoring for farmers to undertake informed decisions for improved crop management.", "date": "2024-07-31", "categories": []}
{"id": "2601.21307", "title": "Mam-App: A Novel Parameter-Efficient Mamba Model for Apple Leaf Disease Classification", "abstract": "The rapid growth of the global population, alongside exponential technological advancement, has intensified the demand for food production. Meeting this demand depends not only on increasing agricultural yield but also on minimizing food loss caused by crop diseases. Diseases account for a substantial portion of apple production losses, despite apples being among the most widely produced and nutritionally valuable fruits worldwide. Previous studies have employed machine learning techniques for feature extraction and early diagnosis of apple leaf diseases, and more recently, deep learning-based models have shown remarkable performance in disease recognition. However, most state-of-the-art deep learning models are highly parameter-intensive, resulting in increased training and inference time. Although lightweight models are more suitable for user-friendly and resource-constrained applications, they often suffer from performance degradation. To address the trade-off between efficiency and performance, we propose Mam-App, a parameter-efficient Mamba-based model for feature extraction and leaf disease classification. The proposed approach achieves competitive state-of-the-art performance on the PlantVillage Apple Leaf Disease dataset, attaining 99.58% accuracy, 99.30% precision, 99.14% recall, and a 99.22% F1-score, while using only 0.051M parameters. This extremely low parameter count makes the model suitable for deployment on drones, mobile devices, and other low-resource platforms. To demonstrate the robustness and generalizability of the proposed model, we further evaluate it on the PlantVillage Corn Leaf Disease and Potato Leaf Disease datasets. The model achieves 99.48%, 99.20%, 99.34%, and 99.27% accuracy, precision, recall, and F1-score on the corn dataset and 98.46%, 98.91%, 95.39%, and 97.01% on the potato dataset, respectively.", "date": "2026-01-29", "categories": []}
{"id": "2411.09693", "title": "CropCraft: Complete Structural Characterization of Crop Plants From Images", "abstract": "The ability to automatically build 3D digital twins of plants from images has countless applications in agriculture, environmental science, robotics, and other fields. However, current 3D reconstruction methods fail to recover complete shapes of plants due to heavy occlusion and complex geometries. In this work, we present a novel method for 3D modeling of agricultural crops based on optimizing a parametric model of plant morphology via inverse procedural modeling. Our method first estimates depth maps by fitting a neural radiance field and then optimizes a specialized loss to estimate morphological parameters that result in consistent depth renderings. The resulting 3D model is complete and biologically plausible. We validate our method on a dataset of real images of agricultural fields, and demonstrate that the reconstructed canopies can be used for a variety of monitoring and simulation applications.", "date": "2024-11-14", "categories": []}
{"id": "2504.14736", "title": "ChronoRoot 2.0: An Open AI-Powered Platform for 2D Temporal Plant Phenotyping", "abstract": "Plant developmental plasticity, particularly in root system architecture, is fundamental to understanding adaptability and agricultural sustainability. ChronoRoot 2.0 builds upon established low-cost hardware while significantly enhancing software capabilities and usability. The system employs nnUNet architecture for multi-class segmentation, demonstrating significant accuracy improvements while simultaneously tracking six distinct plant structures encompassing root, shoot, and seed components: main root, lateral roots, seed, hypocotyl, leaves, and petiole. This architecture enables easy retraining and incorporation of additional training data without requiring machine learning expertise. The platform introduces dual specialized graphical interfaces: a Standard Interface for detailed architectural analysis with novel gravitropic response parameters, and a Screening Interface enabling high-throughput analysis of multiple plants through automated tracking. Functional Principal Component Analysis integration enables discovery of novel phenotypic parameters through temporal pattern comparison. We demonstrate multi-species analysis, with Arabidopsis thaliana and Solanum lycopersicum, both morphologically distinct plant species. Three use cases in Arabidopsis thaliana and validation with tomato seedlings demonstrate enhanced capabilities: circadian growth pattern characterization, gravitropic response analysis in transgenic plants, and high-throughput etiolation screening across multiple genotypes.ChronoRoot 2.0 maintains the low-cost, modular hardware advantages of its predecessor while dramatically improving accessibility through intuitive graphical interfaces and expanded analytical capabilities. The open-source platform makes sophisticated temporal plant phenotyping more accessible to researchers without computational expertise.", "date": "2025-04-20", "categories": []}
{"id": "2601.11612", "title": "Domain-Specific Self-Supervised Pre-training for Agricultural Disease Classification: A Hierarchical Vision Transformer Study", "abstract": "We investigate the impact of domain-specific self-supervised pre-training on agricultural disease classification using hierarchical vision transformers. Our key finding is that SimCLR pre-training on just 3,000 unlabeled agricultural images provides a +4.57% accuracy improvement--exceeding the +3.70% gain from hierarchical architecture design. Critically, we show this SSL benefit is architecture-agnostic: applying the same pre-training to Swin-Base yields +4.08%, to ViT-Base +4.20%, confirming practitioners should prioritize domain data collection over architectural choices. Using HierarchicalViT (HVT), a Swin-style hierarchical transformer, we evaluate on three datasets: Cotton Leaf Disease (7 classes, 90.24%), PlantVillage (38 classes, 96.3%), and PlantDoc (27 classes, 87.1%). At matched parameter counts, HVT-Base (78M) achieves 88.91% vs. Swin-Base (88M) at 87.23%, a +1.68% improvement. For deployment reliability, we report calibration analysis showing HVT achieves 3.56% ECE (1.52% after temperature scaling). Code: https://github.com/w2sg-arnav/HierarchicalViT", "date": "2026-01-09", "categories": []}
{"id": "2601.12366", "title": "DepthCropSeg++: Scaling a Crop Segmentation Foundation Model With Depth-Labeled Data", "abstract": "DepthCropSeg++: a foundation model for crop segmentation, capable of segmenting different crop species under open in-field environment. Crop segmentation is a fundamental task for modern agriculture, which closely relates to many downstream tasks such as plant phenotyping, density estimation, and weed control. In the era of foundation models, a number of generic large language and vision models have been developed. These models have demonstrated remarkable real world generalization due to significant model capacity and largescale datasets. However, current crop segmentation models mostly learn from limited data due to expensive pixel-level labelling cost, often performing well only under specific crop types or controlled environment. In this work, we follow the vein of our previous work DepthCropSeg, an almost unsupervised approach to crop segmentation, to scale up a cross-species and crossscene crop segmentation dataset, with 28,406 images across 30+ species and 15 environmental conditions. We also build upon a state-of-the-art semantic segmentation architecture ViT-Adapter architecture, enhance it with dynamic upsampling for improved detail awareness, and train the model with a two-stage selftraining pipeline. To systematically validate model performance, we conduct comprehensive experiments to justify the effectiveness and generalization capabilities across multiple crop datasets. Results demonstrate that DepthCropSeg++ achieves 93.11% mIoU on a comprehensive testing set, outperforming both supervised baselines and general-purpose vision foundation models like Segmentation Anything Model (SAM) by significant margins (+0.36% and +48.57% respectively). The model particularly excels in challenging scenarios including night-time environment (86.90% mIoU), high-density canopies (90.09% mIoU), and unseen crop varieties (90.09% mIoU), indicating a new state of the art for crop segmentation.", "date": "2026-01-18", "categories": []}
{"id": "2601.13665", "title": "Transformer based Multi-task Fusion Network for Food Spoilage Detection and Shelf life Forecasting", "abstract": "Food wastage is one of the critical challenges in the agricultural supply chain, and accurate and effective spoilage detection can help to reduce it. Further, it is highly important to forecast the spoilage information. This aids the longevity of the supply chain management in the agriculture field. This motivated us to propose fusion based architectures by combining CNN with LSTM and DeiT transformer for the following multi-tasks simultaneously: (i) vegetable classification, (ii) food spoilage detection, and (iii) shelf life forecasting. We developed a dataset by capturing images of vegetables from their fresh state until they were completely spoiled. From the experimental analysis it is concluded that the proposed fusion architectures CNN+CNN-LSTM and CNN+DeiT Transformer outperformed several deep learning models such as CNN, VGG16, ResNet50, Capsule Networks, and DeiT Transformers. Overall, CNN + DeiT Transformer yielded F1-score of 0.98 and 0.61 in vegetable classification and spoilage detection respectively and mean squared error (MSE) and symmetric mean absolute percentage error (SMAPE) of 3.58, and 41.66% respectively in spoilage forecasting. Further, the reliability of the fusion models was validated on noisy images and integrated with LIME to visualize the model decisions.", "date": "2026-01-20", "categories": []}
{"id": "2505.21357", "title": "AgriFM: A Multi-source Temporal Remote Sensing Foundation Model for Agriculture Mapping", "abstract": "Accurate crop mapping fundamentally relies on modeling multi-scale spatiotemporal patterns, where spatial scales range from individual field textures to landscape-level context, and temporal scales capture both short-term phenological transitions and full growing-season dynamics. Transformer-based remote sensing foundation models (RSFMs) offer promising potential for crop mapping due to their innate ability for unified spatiotemporal processing. However, current RSFMs remain suboptimal for crop mapping: they either employ fixed spatiotemporal windows that ignore the multi-scale nature of crop systems or completely disregard temporal information by focusing solely on spatial patterns. To bridge these gaps, we present AgriFM, a multi-source remote sensing foundation model specifically designed for agricultural crop mapping. Our approach begins by establishing the necessity of simultaneous hierarchical spatiotemporal feature extraction, leading to the development of a modified Video Swin Transformer architecture where temporal down-sampling is synchronized with spatial scaling operations. This modified backbone enables efficient unified processing of long time-series satellite inputs. AgriFM leverages temporally rich data streams from three satellite sources including MODIS, Landsat-8/9 and Sentinel-2, and is pre-trained on a global representative dataset comprising over 25 million image samples supervised by land cover products. The resulting framework incorporates a versatile decoder architecture that dynamically fuses these learned spatiotemporal representations, supporting diverse downstream tasks. Comprehensive evaluations demonstrate AgriFM's superior performance over conventional deep learning approaches and state-of-the-art general-purpose RSFMs across all downstream tasks. Codes will be available at https://github.com/flyakon/AgriFM.", "date": "2025-05-27", "categories": []}
{"id": "2505.05513", "title": "Exploring Convolutional Neural Networks for Rice Grain Classification: An Explainable AI Approach", "abstract": "Rice is an essential staple food worldwide that is important in promoting international trade, economic growth, and nutrition. Asian countries such as China, India, Pakistan, Thailand, Vietnam, and Indonesia are notable for their significant contribution to the cultivation and utilization of rice. These nations are also known for cultivating different rice grains, including short and long grains. These sizes are further classified as basmati, jasmine, kainat saila, ipsala, arborio, etc., catering to diverse culinary preferences and cultural traditions. For both local and international trade, inspecting and maintaining the quality of rice grains to satisfy customers and preserve a country's reputation is necessary. Manual quality check and classification is quite a laborious and time-consuming process. It is also highly prone to mistakes. Therefore, an automatic solution must be proposed for the effective and efficient classification of different varieties of rice grains. This research paper presents an automatic framework based on a convolutional neural network (CNN) for classifying different varieties of rice grains. We evaluated the proposed model based on performance metrics such as accuracy, recall, precision, and F1-Score. The CNN model underwent rigorous training and validation, achieving a remarkable accuracy rate and a perfect area under each class's Receiver Operating Characteristic (ROC) curve. The confusion matrix analysis confirmed the model's effectiveness in distinguishing between the different rice varieties, indicating minimal misclassifications. Additionally, the integration of explainability techniques such as LIME (Local Interpretable Model-agnostic Explanations) and SHAP (SHapley Additive exPlanations) provided valuable insights into the model's decision-making process, revealing how specific features of the rice grains influenced classification outcomes.", "date": "2025-05-07", "categories": []}
{"id": "2601.09170", "title": "N-EIoU-YOLOv9: A Signal-Aware Bounding Box Regression Loss for Lightweight Mobile Detection of Rice Leaf Diseases", "abstract": "In this work, we propose N EIoU YOLOv9, a lightweight detection framework based on a signal aware bounding box regression loss derived from non monotonic gradient focusing and geometric decoupling principles, referred to as N EIoU (Non monotonic Efficient Intersection over Union). The proposed loss reshapes localization gradients by combining non monotonic focusing with decoupled width and height optimization, thereby enhancing weak regression signals for hard samples with low overlap while reducing gradient interference. This design is particularly effective for small and low contrast targets commonly observed in agricultural disease imagery. The proposed N EIoU loss is integrated into a lightweight YOLOv9t architecture and evaluated on a self collected field dataset comprising 5908 rice leaf images across four disease categories and healthy leaves. Experimental results demonstrate consistent performance gains over the standard CIoU loss, achieving a mean Average Precision of 90.3 percent, corresponding to a 4.3 percent improvement over the baseline, with improved localization accuracy under stricter evaluation criteria. For practical validation, the optimized model is deployed on an Android device using TensorFlow Lite with Float16 quantization, achieving an average inference time of 156 milliseconds per frame while maintaining accuracy. These results confirm that the proposed approach effectively balances accuracy, optimization stability, and computational efficiency for edge based agricultural monitoring systems.", "date": "2026-01-14", "categories": []}
{"id": "2601.07970", "title": "Sesame Plant Segmentation Dataset: A YOLO Formatted Annotated Dataset", "abstract": "This paper presents the Sesame Plant Segmentation Dataset, an open source annotated image dataset designed to support the development of artificial intelligence models for agricultural applications, with a specific focus on sesame plants. The dataset comprises 206 training images, 43 validation images, and 43 test images in YOLO compatible segmentation format, capturing sesame plants at early growth stages under varying environmental conditions. Data were collected using a high resolution mobile camera from farms in Jirdede, Daura Local Government Area, Katsina State, Nigeria, and annotated using the Segment Anything Model version 2 with farmer supervision. Unlike conventional bounding box datasets, this dataset employs pixel level segmentation to enable more precise detection and analysis of sesame plants in real world farm settings. Model evaluation using the Ultralytics YOLOv8 framework demonstrated strong performance for both detection and segmentation tasks. For bounding box detection, the model achieved a recall of 79 percent, precision of 79 percent, mean average precision at IoU 0.50 of 84 percent, and mean average precision from 0.50 to 0.95 of 58 percent. For segmentation, it achieved a recall of 82 percent, precision of 77 percent, mean average precision at IoU 0.50 of 84 percent, and mean average precision from 0.50 to 0.95 of 52 percent. The dataset represents a novel contribution to sesame focused agricultural vision datasets in Nigeria and supports applications such as plant monitoring, yield estimation, and agricultural research.", "date": "2026-01-12", "categories": []}
{"id": "2601.07975", "title": "An Efficient Additive Kolmogorov-Arnold Transformer for Point-Level Maize Localization in Unmanned Aerial Vehicle Imagery", "abstract": "High-resolution UAV photogrammetry has become a key technology for precision agriculture, enabling centimeter-level crop monitoring and point-level plant localization. However, point-level maize localization in UAV imagery remains challenging due to (1) extremely small object-to-pixel ratios, typically less than 0.1%, (2) prohibitive computational costs of quadratic attention on ultra-high-resolution images larger than 3000 x 4000 pixels, and (3) agricultural scene-specific complexities such as sparse object distribution and environmental variability that are poorly handled by general-purpose vision models.\n  To address these challenges, we propose the Additive Kolmogorov-Arnold Transformer (AKT), which replaces conventional multilayer perceptrons with Pade Kolmogorov-Arnold Network (PKAN) modules to enhance functional expressivity for small-object feature extraction, and introduces PKAN Additive Attention (PAA) to model multiscale spatial dependencies with reduced computational complexity. In addition, we present the Point-based Maize Localization (PML) dataset, consisting of 1,928 high-resolution UAV images with approximately 501,000 point annotations collected under real field conditions.\n  Extensive experiments show that AKT achieves an average F1-score of 62.8%, outperforming state-of-the-art methods by 4.2%, while reducing FLOPs by 12.6% and improving inference throughput by 20.7%. For downstream tasks, AKT attains a mean absolute error of 7.1 in stand counting and a root mean square error of 1.95-1.97 cm in interplant spacing estimation. These results demonstrate that integrating Kolmogorov-Arnold representation theory with efficient attention mechanisms offers an effective framework for high-resolution agricultural remote sensing.", "date": "2026-01-12", "categories": []}
{"id": "2601.03317", "title": "Deep Learning-Based Image Recognition for Soft-Shell Shrimp Classification", "abstract": "With the integration of information technology into aquaculture, production has become more stable and continues to grow annually. As consumer demand for high-quality aquatic products rises, freshness and appearance integrity are key concerns. In shrimp-based processed foods, freshness declines rapidly post-harvest, and soft-shell shrimp often suffer from head-body separation after cooking or freezing, affecting product appearance and consumer perception. To address these issues, this study leverages deep learning-based image recognition for automated classification of white shrimp immediately after harvest. A convolutional neural network (CNN) model replaces manual sorting, enhancing classification accuracy, efficiency, and consistency. By reducing processing time, this technology helps maintain freshness and ensures that shrimp transportation businesses meet customer demands more effectively.", "date": "2026-01-06", "categories": []}
{"id": "2601.03431", "title": "WeedRepFormer: Reparameterizable Vision Transformers for Real-Time Waterhemp Segmentation and Gender Classification", "abstract": "We present WeedRepFormer, a lightweight multi-task Vision Transformer designed for simultaneous waterhemp segmentation and gender classification. Existing agricultural models often struggle to balance the fine-grained feature extraction required for biological attribute classification with the efficiency needed for real-time deployment. To address this, WeedRepFormer systematically integrates structural reparameterization across the entire architecture - comprising a Vision Transformer backbone, a Lite R-ASPP decoder, and a novel reparameterizable classification head - to decouple training-time capacity from inference-time latency. We also introduce a comprehensive waterhemp dataset containing 10,264 annotated frames from 23 plants. On this benchmark, WeedRepFormer achieves 92.18% mIoU for segmentation and 81.91% accuracy for gender classification using only 3.59M parameters and 3.80 GFLOPs. At 108.95 FPS, our model outperforms the state-of-the-art iFormer-T by 4.40% in classification accuracy while maintaining competitive segmentation performance and significantly reducing parameter count by 1.9x.", "date": "2026-01-06", "categories": []}
{"id": "2601.00897", "title": "CornViT: A Multi-Stage Convolutional Vision Transformer Framework for Hierarchical Corn Kernel Analysis", "abstract": "Accurate grading of corn kernels is critical for seed certification, directional seeding, and breeding, yet it is still predominantly performed by manual inspection. This work introduces CornViT, a three-stage Convolutional Vision Transformer (CvT) framework that emulates the hierarchical reasoning of human seed analysts for single-kernel evaluation. Three sequential CvT-13 classifiers operate on 384x384 RGB images: Stage 1 distinguishes pure from impure kernels; Stage 2 categorizes pure kernels into flat and round morphologies; and Stage 3 determines the embryo orientation (up vs. down) for pure, flat kernels. Starting from a public corn seed image collection, we manually relabeled and filtered images to construct three stage-specific datasets: 7265 kernels for purity, 3859 pure kernels for morphology, and 1960 pure-flat kernels for embryo orientation, all released as benchmarks. Head-only fine-tuning of ImageNet-22k pretrained CvT-13 backbones yields test accuracies of 93.76% for purity, 94.11% for shape, and 91.12% for embryo-orientation detection. Under identical training conditions, ResNet-50 reaches only 76.56 to 81.02 percent, whereas DenseNet-121 attains 86.56 to 89.38 percent accuracy. These results highlight the advantages of convolution-augmented self-attention for kernel analysis. To facilitate adoption, we deploy CornViT in a Flask-based web application that performs stage-wise inference and exposes interpretable outputs through a browser interface. Together, the CornViT framework, curated datasets, and web application provide a deployable solution for automated corn kernel quality assessment in seed quality workflows. Source code and data are publicly available.", "date": "2025-12-31", "categories": []}
{"id": "2601.01084", "title": "A UAV-Based Multispectral and RGB Dataset for Multi-Stage Paddy Crop Monitoring in Indian Agricultural Fields", "abstract": "We present a large-scale unmanned aerial vehicle (UAV)-based RGB and multispectral image dataset collected over paddy fields in the Vijayawada region, Andhra Pradesh, India, covering nursery to harvesting stages. We used a 20-megapixel RGB camera and a 5-megapixel four-band multispectral camera capturing red, green, red-edge, and near-infrared bands. Standardised operating procedure (SOP) and checklists were developed to ensure repeatable data acquisition. Our dataset comprises of 42,430 raw images (415 GB) captured over 5 acres with 1 cm/pixel ground sampling distance (GSD) with associated metadata such as GPS coordinates, flight altitude, and environmental conditions. Captured images were validated using Pix4D Fields to generate orthomosaic maps and vegetation index maps, such as normalised difference vegetation index (NDVI) and normalised difference red-edge (NDRE) index. Our dataset is one of the few datasets that provide high-resolution images with rich metadata that cover all growth stages of Indian paddy crops. The dataset is available on IEEE DataPort with DOI, . It can support studies on targeted spraying, disease analysis, and yield estimation.", "date": "2026-01-03", "categories": []}
{"id": "2512.22990", "title": "A Low-Cost UAV Deep Learning Pipeline for Integrated Apple Disease Diagnosis,Freshness Assessment, and Fruit Detection", "abstract": "Apple orchards require timely disease detection, fruit quality assessment, and yield estimation, yet existing UAV-based systems address such tasks in isolation and often rely on costly multispectral sensors. This paper presents a unified, low-cost RGB-only UAV-based orchard intelligent pipeline integrating ResNet50 for leaf disease detection, VGG 16 for apple freshness determination, and YOLOv8 for real-time apple detection and localization. The system runs on an ESP32-CAM and Raspberry Pi, providing fully offline on-site inference without cloud support. Experiments demonstrate 98.9% accuracy for leaf disease classification, 97.4% accuracy for freshness classification, and 0.857 F1 score for apple detection. The framework provides an accessible and scalable alternative to multispectral UAV solutions, supporting practical precision agriculture on affordable hardware.", "date": "2025-12-28", "categories": []}
{"id": "2601.00243", "title": "Context-Aware Pesticide Recommendation via Few-Shot Pest Recognition for Precision Agriculture", "abstract": "Effective pest management is crucial for enhancing agricultural productivity, especially for crops such as sugarcane and wheat that are highly vulnerable to pest infestations. Traditional pest management methods depend heavily on manual field inspections and the use of chemical pesticides. These approaches are often costly, time-consuming, labor-intensive, and can have a negative impact on the environment. To overcome these challenges, this study presents a lightweight framework for pest detection and pesticide recommendation, designed for low-resource devices such as smartphones and drones, making it suitable for use by small and marginal farmers.\n  The proposed framework includes two main components. The first is a Pest Detection Module that uses a compact, lightweight convolutional neural network (CNN) combined with prototypical meta-learning to accurately identify pests even when only a few training samples are available. The second is a Pesticide Recommendation Module that incorporates environmental factors like crop type and growth stage to suggest safe and eco-friendly pesticide recommendations. To train and evaluate our framework, a comprehensive pest image dataset was developed by combining multiple publicly available datasets. The final dataset contains samples with different viewing angles, pest sizes, and background conditions to ensure strong generalization.\n  Experimental results show that the proposed lightweight CNN achieves high accuracy, comparable to state-of-the-art models, while significantly reducing computational complexity. The Decision Support System additionally improves pest management by reducing dependence on traditional chemical pesticides and encouraging sustainable practices, demonstrating its potential for real-time applications in precision agriculture.", "date": "2026-01-01", "categories": []}
{"id": "2512.24193", "title": "PointRAFT: 3D deep learning for high-throughput prediction of potato tuber weight from partial point clouds", "abstract": "Potato yield is a key indicator for optimizing cultivation practices in agriculture. Potato yield can be estimated on harvesters using RGB-D cameras, which capture three-dimensional (3D) information of individual tubers moving along the conveyor belt. However, point clouds reconstructed from RGB-D images are incomplete due to self-occlusion, leading to systematic underestimation of tuber weight. To address this, we introduce PointRAFT, a high-throughput point cloud regression network that directly predicts continuous 3D shape properties, such as tuber weight, from partial point clouds. Rather than reconstructing full 3D geometry, PointRAFT infers target values directly from raw 3D data. Its key architectural novelty is an object height embedding that incorporates tuber height as an additional geometric cue, improving weight prediction under practical harvesting conditions. PointRAFT was trained and evaluated on 26,688 partial point clouds collected from 859 potato tubers across four cultivars and three growing seasons on an operational harvester in Japan. On a test set of 5,254 point clouds from 172 tubers, PointRAFT achieved a mean absolute error of 12.0 g and a root mean squared error of 17.2 g, substantially outperforming a linear regression baseline and a standard PointNet++ regression network. With an average inference time of 6.3 ms per point cloud, PointRAFT supports processing rates of up to 150 tubers per second, meeting the high-throughput requirements of commercial potato harvesters. Beyond potato weight estimation, PointRAFT provides a versatile regression network applicable to a wide range of 3D phenotyping and robotic perception tasks. The code, network weights, and a subset of the dataset are publicly available at https://github.com/pieterblok/pointraft.git.", "date": "2025-12-30", "categories": []}
{"id": "2512.24947", "title": "CPJ: Explainable Agricultural Pest Diagnosis via Caption-Prompt-Judge with LLM-Judged Refinement", "abstract": "Accurate and interpretable crop disease diagnosis is essential for agricultural decision-making, yet existing methods often rely on costly supervised fine-tuning and perform poorly under domain shifts. We propose Caption--Prompt--Judge (CPJ), a training-free few-shot framework that enhances Agri-Pest VQA through structured, interpretable image captions. CPJ employs large vision-language models to generate multi-angle captions, refined iteratively via an LLM-as-Judge module, which then inform a dual-answer VQA process for both recognition and management responses. Evaluated on CDDMBench, CPJ significantly improves performance: using GPT-5-mini captions, GPT-5-Nano achieves \\textbf{+22.7} pp in disease classification and \\textbf{+19.5} points in QA score over no-caption baselines. The framework provides transparent, evidence-based reasoning, advancing robust and explainable agricultural diagnosis without fine-tuning. Our code and data are publicly available at: https://github.com/CPJ-Agricultural/CPJ-Agricultural-Diagnosis.", "date": "2025-12-31", "categories": []}
{"id": "2506.18060", "title": "Fine-Tuned Vision Transformers Capture Complex Wheat Spike Morphology for Volume Estimation from RGB Images", "abstract": "Estimating three-dimensional morphological traits such as volume from two-dimensional RGB images presents inherent challenges due to the loss of depth information, projection distortions, and occlusions under field conditions. In this work, we explore multiple approaches for non-destructive volume estimation of wheat spikes using RGB images and structured-light 3D scans as ground truth references. Wheat spike volume is promising for phenotyping as it shows high correlation with spike dry weight, a key component of fruiting efficiency. Accounting for the complex geometry of the spikes, we compare different neural network approaches for volume estimation from 2D images and benchmark them against two conventional baselines: a 2D area-based projection and a geometric reconstruction using axis-aligned cross-sections. Fine-tuned Vision Transformers (DINOv2 and DINOv3) with MLPs achieve the lowest MAPE of 5.08\\% and 4.67\\% and the highest correlation of 0.96 and 0.97 on six-view indoor images, outperforming fine-tuned CNNs (ResNet18 and ResNet50), wheat-specific backbones, and both baselines. When using frozen DINO backbones, deep-supervised LSTMs outperform MLPs, whereas after fine-tuning, improved high-level representations allow simple MLPs to outperform LSTMs. We demonstrate that object shape significantly impacts volume estimation accuracy, with irregular geometries such as wheat spikes posing greater challenges for geometric methods than for deep learning approaches. Fine-tuning DINOv3 on field-based single side-view images yields a MAPE of 8.39\\% and a correlation of 0.90, providing a novel pipeline and a fast, accurate, and non-destructive approach for wheat spike volume phenotyping.", "date": "2025-06-22", "categories": []}
{"id": "2303.05634", "title": "Fusarium head blight detection, spikelet estimation, and severity assessment in wheat using 3D convolutional neural networks", "abstract": "Fusarium head blight (FHB) is one of the most significant diseases affecting wheat and other small grain cereals worldwide. The development of resistant varieties requires the laborious task of field and greenhouse phenotyping. The applications considered in this work are the automated detection of FHB disease symptoms expressed on a wheat plant, the automated estimation of the total number of spikelets and the total number of infected spikelets on a wheat head, and the automated assessment of the FHB severity in infected wheat. The data used to generate the results are 3-dimensional (3D) multispectral point clouds (PC), which are 3D collections of points - each associated with a red, green, blue (RGB), and near-infrared (NIR) measurement. Over 300 wheat plant images were collected using a multispectral 3D scanner, and the labelled UW-MRDC 3D wheat dataset was created. The data was used to develop novel and efficient 3D convolutional neural network (CNN) models for FHB detection, which achieved 100% accuracy. The influence of the multispectral information on performance was evaluated, and our results showed the dominance of the RGB channels over both the NIR and the NIR plus RGB channels combined. Furthermore, novel and efficient 3D CNNs were created to estimate the total number of spikelets and the total number of infected spikelets on a wheat head, and our best models achieved mean absolute errors (MAE) of 1.13 and 1.56, respectively. Moreover, 3D CNN models for FHB severity estimation were created, and our best model achieved 8.6 MAE. A linear regression analysis between the visual FHB severity assessment and the FHB severity predicted by our 3D CNN was performed, and the results showed a significant correlation between the two variables with a 0.0001 P-value and 0.94 R-squared.", "date": "2023-03-10", "categories": []}
{"id": "2512.19989", "title": "A Novel CNN Gradient Boosting Ensemble for Guava Disease Detection", "abstract": "As a significant agricultural country, Bangladesh utilizes its fertile land for guava cultivation and dedicated labor to boost its economic development. In a nation like Bangladesh, enhancing guava production and agricultural practices plays a crucial role in its economy. Anthracnose and fruit fly infection can lower the quality and productivity of guava, a crucial tropical fruit. Expert systems that detect diseases early can reduce losses and safeguard the harvest. Images of guava fruits classified into the Healthy, Fruit Flies, and Anthracnose classes are included in the Guava Fruit Disease Dataset 2024 (GFDD24), which comes from plantations in Rajshahi and Pabna, Bangladesh. This study aims to create models using CNN alongside traditional machine learning techniques that can effectively identify guava diseases in locally cultivated varieties in Bangladesh. In order to achieve the highest classification accuracy of approximately 99.99% for the guava dataset, we propose utilizing ensemble models that combine CNNML with Gradient Boosting Machine. In general, the CNN-ML cascade framework exhibits strong, high-accuracy guava disease detection that is appropriate for real-time agricultural monitoring systems.", "date": "2025-12-23", "categories": []}
{"id": "2512.20148", "title": "Enhancing annotations for 5D apple pose estimation through 3D Gaussian Splatting (3DGS)", "abstract": "Automating tasks in orchards is challenging because of the large amount of variation in the environment and occlusions. One of the challenges is apple pose estimation, where key points, such as the calyx, are often occluded. Recently developed pose estimation methods no longer rely on these key points, but still require them for annotations, making annotating challenging and time-consuming. Due to the abovementioned occlusions, there can be conflicting and missing annotations of the same fruit between different images. Novel 3D reconstruction methods can be used to simplify annotating and enlarge datasets. We propose a novel pipeline consisting of 3D Gaussian Splatting to reconstruct an orchard scene, simplified annotations, automated projection of the annotations to images, and the training and evaluation of a pose estimation method. Using our pipeline, 105 manual annotations were required to obtain 28,191 training labels, a reduction of 99.6%. Experimental results indicated that training with labels of fruits that are $\\leq95\\%$ occluded resulted in the best performance, with a neutral F1 score of 0.927 on the original images and 0.970 on the rendered images. Adjusting the size of the training dataset had small effects on the model performance in terms of F1 score and pose estimation accuracy. It was found that the least occluded fruits had the best position estimation, which worsened as the fruits became more occluded. It was also found that the tested pose estimation method was unable to correctly learn the orientation estimation of apples.", "date": "2025-12-23", "categories": []}
{"id": "2512.18344", "title": "MCVI-SANet: A lightweight semi-supervised model for LAI and SPAD estimation of winter wheat under vegetation index saturation", "abstract": "Vegetation index (VI) saturation during the dense canopy stage and limited ground-truth annotations of winter wheat constrain accurate estimation of LAI and SPAD. Existing VI-based and texture-driven machine learning methods exhibit limited feature expressiveness. In addition, deep learning baselines suffer from domain gaps and high data demands, which restrict their generalization. Therefore, this study proposes the Multi-Channel Vegetation Indices Saturation Aware Net (MCVI-SANet), a lightweight semi-supervised vision model. The model incorporates a newly designed Vegetation Index Saturation-Aware Block (VI-SABlock) for adaptive channel-spatial feature enhancement. It also integrates a VICReg-based semi-supervised strategy to further improve generalization. Datasets were partitioned using a vegetation height-informed strategy to maintain representativeness across growth stages. Experiments over 10 repeated runs demonstrate that MCVI-SANet achieves state-of-the-art accuracy. The model attains an average R2 of 0.8123 and RMSE of 0.4796 for LAI, and an average R2 of 0.6846 and RMSE of 2.4222 for SPAD. This performance surpasses the best-performing baselines, with improvements of 8.95% in average LAI R2 and 8.17% in average SPAD R2. Moreover, MCVI-SANet maintains high inference speed with only 0.10M parameters. Overall, the integration of semi-supervised learning with agronomic priors provides a promising approach for enhancing remote sensing-based precision agriculture.", "date": "2025-12-20", "categories": []}
{"id": "2502.13764", "title": "An Overall Real-Time Mechanism for Classification and Quality Evaluation of Rice", "abstract": "Rice is one of the most widely cultivated crops globally and has been developed into numerous varieties. The quality of rice during cultivation is primarily determined by its cultivar and characteristics. Traditionally, rice classification and quality assessment rely on manual visual inspection, a process that is both time-consuming and prone to errors. However, with advancements in machine vision technology, automating rice classification and quality evaluation based on its cultivar and characteristics has become increasingly feasible, enhancing both accuracy and efficiency. This study proposes a real-time evaluation mechanism for comprehensive rice grain assessment, integrating a one-stage object detection approach, a deep convolutional neural network, and traditional machine learning techniques. The proposed framework enables rice variety identification, grain completeness grading, and grain chalkiness evaluation. The rice grain dataset used in this study comprises approximately 20,000 images from six widely cultivated rice varieties in China. Experimental results demonstrate that the proposed mechanism achieves a mean average precision (mAP) of 99.14% in the object detection task and an accuracy of 97.89% in the classification task. Furthermore, the framework attains an average accuracy of 97.56% in grain completeness grading within the same rice variety, contributing to an effective quality evaluation system.", "date": "2025-02-19", "categories": []}
{"id": "2503.03111", "title": "An Improved Pure Fully Connected Neural Network for Rice Grain Classification", "abstract": "Rice is a staple food for a significant portion of the world's population, providing essential nutrients and serving as a versatile in-gredient in a wide range of culinary traditions. Recently, the use of deep learning has enabled automated classification of rice, im-proving accuracy and efficiency. However, classical models based on first-stage training may face difficulties in distinguishing between rice varieties with similar external characteristics, thus leading to misclassifications. Considering the transparency and feasibility of model, we selected and gradually improved pure fully connected neural network to achieve classification of rice grain. The dataset we used contains both global and domestic rice images obtained from websites and laboratories respectively. First, the training mode was changed from one-stage training to two-stage training, which significantly contributes to distinguishing two similar types of rice. Secondly, the preprocessing method was changed from random tilting to horizontal or vertical position cor-rection. After those two enhancements, the accuracy of our model increased notably from 97% to 99%. In summary, two subtle methods proposed in this study can remarkably enhance the classification ability of deep learning models in terms of the classification of rice grain.", "date": "2025-03-05", "categories": []}
{"id": "2504.21706", "title": "Vision Transformers in Precision Agriculture: A Comprehensive Survey", "abstract": "Detecting plant diseases is a crucial aspect of modern agriculture, as it plays a key role in maintaining crop health and increasing overall yield. Traditional approaches, though still valuable, often rely on manual inspection or conventional machine learning techniques, both of which face limitations in scalability and accuracy. Recently, Vision Transformers (ViTs) have emerged as a promising alternative, offering advantages such as improved handling of long-range dependencies and better scalability for visual tasks. This review explores the application of ViTs in precision agriculture, covering a range of tasks. We begin by introducing the foundational architecture of ViTs and discussing their transition from Natural Language Processing (NLP) to Computer Vision. The discussion includes the concept of inductive bias in traditional models like Convolutional Neural Networks (CNNs), and how ViTs mitigate these biases. We provide a comprehensive review of recent literature, focusing on key methodologies, datasets, and performance metrics. This study also includes a comparative analysis of CNNs and ViTs, along with a review of hybrid models and performance enhancements. Technical challenges such as data requirements, computational demands, and model interpretability are addressed, along with potential solutions. Finally, we outline future research directions and technological advancements that could further support the integration of ViTs in real-world agricultural settings. Our goal with this study is to offer practitioners and researchers a deeper understanding of how ViTs are poised to transform smart and precision agriculture.", "date": "2025-04-30", "categories": []}
{"id": "2512.12357", "title": "TCLeaf-Net: a transformer-convolution framework with global-local attention for robust in-field lesion-level plant leaf disease detection", "abstract": "Timely and accurate detection of foliar diseases is vital for safeguarding crop growth and reducing yield losses. Yet, in real-field conditions, cluttered backgrounds, domain shifts, and limited lesion-level datasets hinder robust modeling. To address these challenges, we release Daylily-Leaf, a paired lesion-level dataset comprising 1,746 RGB images and 7,839 lesions captured under both ideal and in-field conditions, and propose TCLeaf-Net, a transformer-convolution hybrid detector optimized for real-field use. TCLeaf-Net is designed to tackle three major challenges. To mitigate interference from complex backgrounds, the transformer-convolution module (TCM) couples global context with locality-preserving convolution to suppress non-leaf regions. To reduce information loss during downsampling, the raw-scale feature recalling and sampling (RSFRS) block combines bilinear resampling and convolution to preserve fine spatial detail. To handle variations in lesion scale and feature shifts, the deformable alignment block with FPN (DFPN) employs offset-based alignment and multi-receptive-field perception to strengthen multi-scale fusion. Experimental results show that on the in-field split of the Daylily-Leaf dataset, TCLeaf-Net improves mAP@50 by 5.4 percentage points over the baseline model, reaching 78.2\\%, while reducing computation by 7.5 GFLOPs and GPU memory usage by 8.7\\%. Moreover, the model outperforms recent YOLO and RT-DETR series in both precision and recall, and demonstrates strong performance on the PlantDoc, Tomato-Leaf, and Rice-Leaf datasets, validating its robustness and generalizability to other plant disease detection scenarios.", "date": "2025-12-13", "categories": []}
{"id": "2512.13641", "title": "From Code to Field: Evaluating the Robustness of Convolutional Neural Networks for Disease Diagnosis in Mango Leaves", "abstract": "The validation and verification of artificial intelligence (AI) models through robustness assessment are essential to guarantee the reliable performance of intelligent systems facing real-world challenges, such as image corruptions including noise, blurring, and weather variations. Despite the global importance of mango (Mangifera indica L.), there is a lack of studies on the robustness of models for the diagnosis of disease in its leaves. This paper proposes a methodology to evaluate convolutional neural networks (CNNs) under adverse conditions. We adapted the MangoLeafDB dataset, generating MangoLeafDB-C with 19 types of artificial corruptions at five severity levels. We conducted a benchmark comparing five architectures: ResNet-50, ResNet-101, VGG-16, Xception, and LCNN (the latter being a lightweight architecture designed specifically for mango leaf diagnosis). The metrics include the F1 score, the corruption error (CE) and the relative mean corruption error (relative mCE). The results show that LCNN outperformed complex models in corruptions that can be present in real-world scenarios such as Defocus Blur, Motion Blur, while also achieving the lowest mCE. Modern architectures (e.g., ResNet-101) exhibited significant performance degradation in corrupted scenarios, despite their high accuracy under ideal conditions. These findings suggest that lightweight and specialized models may be more suitable for real-world applications in edge devices, where robustness and efficiency are critical. The study highlights the need to incorporate robustness assessments in the development of intelligent systems for agriculture, particularly in regions with technological limitations.", "date": "2025-12-15", "categories": []}
{"id": "2512.11360", "title": "Reliable Detection of Minute Targets in High-Resolution Aerial Imagery across Temporal Shifts", "abstract": "Efficient crop detection via Unmanned Aerial Vehicles is critical for scaling precision agriculture, yet it remains challenging due to the small scale of targets and environmental variability. This paper addresses the detection of rice seedlings in paddy fields by leveraging a Faster R-CNN architecture initialized via transfer learning. To overcome the specific difficulties of detecting minute objects in high-resolution aerial imagery, we curate a significant UAV dataset for training and rigorously evaluate the model's generalization capabilities. Specifically, we validate performance across three distinct test sets acquired at different temporal intervals, thereby assessing robustness against varying imaging conditions. Our empirical results demonstrate that transfer learning not only facilitates the rapid convergence of object detection models in agricultural contexts but also yields consistent performance despite domain shifts in image acquisition.", "date": "2025-12-12", "categories": []}
{"id": "2506.22438", "title": "Counting with Confidence: Accurate Pest Monitoring in Water Traps", "abstract": "Accurate pest population monitoring and tracking their dynamic changes are crucial for precision agriculture decision-making. A common limitation in existing vision-based automatic pest counting research is that models are typically evaluated on datasets with ground truth but deployed in real-world scenarios without assessing the reliability of counting results due to the lack of ground truth. To this end, this paper proposed a method for comprehensively evaluating pest counting confidence in the image, based on information related to counting results and external environmental conditions. First, a pest detection network is used for pest detection and counting, extracting counting result-related information. Then, the pest images undergo image quality assessment, image complexity assessment, and pest distribution uniformity assessment. And the changes in image clarity caused by stirring during image acquisition are quantified by calculating the average gradient magnitude. Notably, we designed a hypothesis-driven multi-factor sensitivity analysis method to select the optimal image quality assessment and image complexity assessment methods. And we proposed an adaptive DBSCAN clustering algorithm for pest distribution uniformity assessment. Finally, the obtained information related to counting results and external environmental conditions is input into a regression model for prediction, resulting in the final pest counting confidence. To the best of our knowledge, this is the first study dedicated to comprehensively evaluating counting confidence in counting tasks, and quantifying the relationship between influencing factors and counting confidence through a model. Experimental results show our method reduces MSE by 31.7% and improves R2 by 15.2% on the pest counting confidence test set, compared to the baseline built primarily on information related to counting results.", "date": "2025-05-19", "categories": []}
{"id": "2512.08747", "title": "A Scalable Pipeline Combining Procedural 3D Graphics and Guided Diffusion for Photorealistic Synthetic Training Data Generation in White Button Mushroom Segmentation", "abstract": "Industrial mushroom cultivation increasingly relies on computer vision for monitoring and automated harvesting. However, developing accurate detection and segmentation models requires large, precisely annotated datasets that are costly to produce. Synthetic data provides a scalable alternative, yet often lacks sufficient realism to generalize to real-world scenarios. This paper presents a novel workflow that integrates 3D rendering in Blender with a constrained diffusion model to automatically generate high-quality annotated, photorealistic synthetic images of Agaricus Bisporus mushrooms. This approach preserves full control over 3D scene configuration and annotations while achieving photorealism without the need for specialized computer graphics expertise. We release two synthetic datasets (each containing 6,000 images depicting over 250k mushroom instances) and evaluate Mask R-CNN models trained on them in a zero-shot setting. When tested on two independent real-world datasets (including a newly collected benchmark), our method achieves state-of-the-art segmentation performance (F1 = 0.859 on M18K), despite using only synthetic training data. Although the approach is demonstrated on Agaricus Bisporus mushrooms, the proposed pipeline can be readily adapted to other mushroom species or to other agricultural domains, such as fruit and leaf detection.", "date": "2025-12-09", "categories": []}
{"id": "2510.04002", "title": "AgriGPT-VL: Agricultural Vision-Language Understanding Suite", "abstract": "Despite rapid advances in multimodal large language models, agricultural applications remain constrained by the scarcity of domain-tailored models, curated vision-language corpora, and rigorous evaluation. To address these challenges, we present the AgriGPT-VL Suite, a unified multimodal framework for agriculture. Our contributions are threefold. First, we introduce Agri-3M-VL, the largest vision-language corpus for agriculture to our knowledge, curated by a scalable multi-agent data generator; it comprises 1M image-caption pairs, 2M image-grounded VQA pairs, 50K expert-level VQA instances, and 15K GRPO reinforcement learning samples. Second, we develop AgriGPT-VL, an agriculture-specialized vision-language model trained via a progressive curriculum of textual grounding, multimodal shallow/deep alignment, and GRPO refinement. This method achieves strong multimodal reasoning while preserving text-only capability. Third, we establish AgriBench-VL-4K, a compact yet challenging evaluation suite with open-ended and image-grounded questions, paired with multi-metric evaluation and an LLM-as-a-judge framework. Experiments show that AgriGPT-VL outperforms leading general-purpose VLMs on AgriBench-VL-4K, achieving higher pairwise win rates in the LLM-as-a-judge evaluation. Meanwhile, it remains competitive on the text-only AgriBench-13K with no noticeable degradation of language ability. Ablation studies further confirm consistent gains from our alignment and GRPO refinement stages. We will open source all of the resources to support reproducible research and deployment in low-resource agricultural settings.", "date": "2025-10-05", "categories": []}
{"id": "2407.21150", "title": "PLANesT-3D: A new annotated dataset for segmentation of 3D plant point clouds", "abstract": "Creation of new annotated public datasets is crucial in helping advances in 3D computer vision and machine learning meet their full potential for automatic interpretation of 3D plant models. Despite the proliferation of deep neural network architectures for segmentation and phenotyping of 3D plant models in the last decade, the amount of data, and diversity in terms of species and data acquisition modalities are far from sufficient for evaluation of such tools for their generalization ability. To contribute to closing this gap, we introduce PLANesT-3D; a new annotated dataset of 3D color point clouds of plants. PLANesT-3D is composed of 34 point cloud models representing 34 real plants from three different plant species: \\textit{Capsicum annuum}, \\textit{Rosa kordana}, and \\textit{Ribes rubrum}. Both semantic labels in terms of \"leaf\" and \"stem\", and organ instance labels were manually annotated for the full point clouds. PLANesT-3D introduces diversity to existing datasets by adding point clouds of two new species and providing 3D data acquired with the low-cost SfM/MVS technique as opposed to laser scanning or expensive setups. Point clouds reconstructed with SfM/MVS modality exhibit challenges such as missing data, variable density, and illumination variations. As an additional contribution, SP-LSCnet, a novel semantic segmentation method that is a combination of unsupervised superpoint extraction and a 3D point-based deep learning approach is introduced and evaluated on the new dataset. The advantages of SP-LSCnet over other deep learning methods are its modular structure and increased interpretability. Two existing deep neural network architectures, PointNet++ and RoseSegNet, were also tested on the point clouds of PLANesT-3D for semantic segmentation.", "date": "2024-07-30", "categories": []}
{"id": "2512.05412", "title": "YOLO and SGBM Integration for Autonomous Tree Branch Detection and Depth Estimation in Radiata Pine Pruning Applications", "abstract": "Manual pruning of radiata pine trees poses significant safety risks due to extreme working heights and challenging terrain. This paper presents a computer vision framework that integrates YOLO object detection with Semi-Global Block Matching (SGBM) stereo vision for autonomous drone-based pruning operations. Our system achieves precise branch detection and depth estimation using only stereo camera input, eliminating the need for expensive LiDAR sensors. Experimental evaluation demonstrates YOLO's superior performance over Mask R-CNN, achieving 82.0% mAPmask50-95 for branch segmentation. The integrated system accurately localizes branches within a 2 m operational range, with processing times under one second per frame. These results establish the feasibility of cost-effective autonomous pruning systems that enhance worker safety and operational efficiency in commercial forestry.", "date": "2025-12-05", "categories": []}
{"id": "2511.23222", "title": "DAONet-YOLOv8: An Occlusion-Aware Dual-Attention Network for Tea Leaf Pest and Disease Detection", "abstract": "Accurate detection of tea leaf pests and diseases in real plantations remains challenging due to complex backgrounds, variable illumination, and frequent occlusions among dense branches and leaves. Existing detectors often suffer from missed detections and false positives in such scenarios. To address these issues, we propose DAONet-YOLOv8, an enhanced YOLOv8 variant with three key improvements: (1) a Dual-Attention Fusion Module (DAFM) that combines convolutional local feature extraction with self-attention based global context modeling to focus on subtle lesion regions while suppressing background noise; (2) an occlusion-aware detection head (Detect-OAHead) that learns the relationship between visible and occluded parts to compensate for missing lesion features; and (3) a C2f-DSConv module employing dynamic synthesis convolutions with multiple kernel shapes to better capture irregular lesion boundaries. Experiments on our real-world tea plantation dataset containing six pest and disease categories demonstrate that DAONet-YOLOv8 achieves 92.97% precision, 92.80% recall, 97.10% mAP@50 and 76.90% mAP@50:95, outperforming the YOLOv8n baseline by 2.34, 4.68, 1.40 and 1.80 percentage points respectively, while reducing parameters by 16.7%. Comparative experiments further confirm that DAONet-YOLOv8 achieves superior performance over mainstream detection models.", "date": "2025-11-28", "categories": []}
{"id": "2509.12247", "title": "Modular, On-Site Solutions with Lightweight Anomaly Detection for Sustainable Nutrient Management in Agriculture", "abstract": "Efficient nutrient management is critical for crop growth and sustainable resource consumption (e.g., nitrogen, energy). Current approaches require lengthy analyses, preventing real-time optimization; similarly, imaging facilitates rapid phenotyping but can be computationally intensive, preventing deployment under resource constraints. This study proposes a flexible, tiered pipeline for anomaly detection and status estimation (fresh weight, dry mass, and tissue nutrients), including a comprehensive energy analysis of approaches that span the efficiency-accuracy spectrum. Using a nutrient depletion experiment with three treatments (T1-100%, T2-50%, and T3-25% fertilizer strength) and multispectral imaging (MSI), we developed a hierarchical pipeline using an autoencoder (AE) for early warning. Further, we compared two status estimation modules of different complexity for more detailed analysis: vegetation index (VI) features with machine learning (Random Forest, RF) and raw whole-image deep learning (Vision Transformer, ViT). Results demonstrated high-efficiency anomaly detection (73% net detection of T3 samples 9 days after transplanting) at substantially lower energy than embodied energy in wasted nitrogen. The state estimation modules show trade-offs, with ViT outperforming RF on phosphorus and calcium estimation (R2 0.61 vs. 0.58, 0.48 vs. 0.35) at higher energy cost. With our modular pipeline, this work opens opportunities for edge diagnostics and practical opportunities for agricultural sustainability.", "date": "2025-09-11", "categories": []}
{"id": "2511.11659", "title": "DWFF-Net : A Multi-Scale Farmland System Habitat Identification Method with Adaptive Dynamic Weight", "abstract": "Addressing the current lack of a standardized habitat classification system for cultivated land ecosystems, incomplete coverage of the habitat types, and the inability of existing models to effectively integrate semantic and texture features-resulting in insufficient segmentation accuracy and blurred boundaries for multi-scale habitats (e.g., large-scale field plots and micro-habitats)-this study developed a comprehensively annotated ultra-high-resolution remote sensing image dataset encompassing 15 categories of cultivated land system habitats. Furthermore, we propose a Dynamic-Weighted Feature Fusion Network (DWFF-Net). The encoder of this model utilizes a frozen-parameter DINOv3 to extract foundational features. By analyzing the relationships between different category images and feature maps, we introduce a data-level adaptive dynamic weighting strategy for feature fusion. The decoder incorporates a dynamic weight computation network to achieve thorough integration of multi-layer features, and a hybrid loss function is adopted to optimize model training. Experimental results on the constructed dataset demonstrate that the proposed model achieves a mean Intersection over Union (mIoU) of 69.79% and an F1-score of 80.49%, outperforming the baseline network by 2.1% and 1.61%, respectively. Ablation studies further confirm the complementary nature of multi-layer feature fusion, which effectively improves the IoU for micro-habitat categories such as field ridges. This study establishes a habitat identification framework for cultivated land systems based on adaptive multi-layer feature fusion, enabling sub-meter precision habitat mapping at a low cost and providing robust technical support for fine-grained habitat monitoring in cultivated landscapes. (The complete code repository can be accessed via GitHub at the following URL: https://github.com/sysau/DWFF-Net)", "date": "2025-11-11", "categories": []}
{"id": "2506.03213", "title": "ConMamba: Contrastive Vision Mamba for Plant Disease Detection", "abstract": "Plant Disease Detection (PDD) is a key aspect of precision agriculture. However, existing deep learning methods often rely on extensively annotated datasets, which are time-consuming and costly to generate. Self-supervised Learning (SSL) offers a promising alternative by exploiting the abundance of unlabeled data. However, most existing SSL approaches suffer from high computational costs due to convolutional neural networks or transformer-based architectures. Additionally, they struggle to capture long-range dependencies in visual representation and rely on static loss functions that fail to align local and global features effectively. To address these challenges, we propose ConMamba, a novel SSL framework specially designed for PDD. ConMamba integrates the Vision Mamba Encoder (VME), which employs a bidirectional State Space Model (SSM) to capture long-range dependencies efficiently. Furthermore, we introduce a dual-level contrastive loss with dynamic weight adjustment to optimize local-global feature alignment. Experimental results on three benchmark datasets demonstrate that ConMamba significantly outperforms state-of-the-art methods across multiple evaluation metrics. This provides an efficient and robust solution for PDD.", "date": "2025-06-03", "categories": []}
{"id": "2511.16982", "title": "A Diversity-optimized Deep Ensemble Approach for Accurate Plant Leaf Disease Detection", "abstract": "Plant diseases pose a significant threat to global agriculture, causing over $220 billion in annual economic losses and jeopardizing food security. The timely and accurate detection of these diseases from plant leaf images is critical to mitigating their adverse effects. Deep neural network Ensembles (Deep Ensembles) have emerged as a powerful approach to enhancing prediction accuracy by leveraging the strengths of diverse Deep Neural Networks (DNNs). However, selecting high-performing ensemble member models is challenging due to the inherent difficulty in measuring ensemble diversity. In this paper, we introduce the Synergistic Diversity (SQ) framework to enhance plant disease detection accuracy. First, we conduct a comprehensive analysis of the limitations of existing ensemble diversity metrics (denoted as Q metrics), which often fail to identify optimal ensemble teams. Second, we present the SQ metric, a novel measure that captures the synergy between ensemble members and consistently aligns with ensemble accuracy. Third, we validate our SQ approach through extensive experiments on a plant leaf image dataset, which demonstrates that our SQ metric substantially improves ensemble selection and enhances detection accuracy. Our findings pave the way for a more reliable and efficient image-based plant disease detection.", "date": "2025-11-21", "categories": []}
{"id": "2511.16294", "title": "Explainable AI for Diabetic Retinopathy Detection Using Deep Learning with Attention Mechanisms and Fuzzy Logic-Based Interpretability", "abstract": "The task of weed detection is an essential element of precision agriculture since accurate species identification allows a farmer to selectively apply herbicides and fits into sustainable agriculture crop management. This paper proposes a hybrid deep learning framework recipe for weed detection that utilizes Convolutional Neural Networks (CNNs), Vision Transformers (ViTs), and Graph Neural Networks (GNNs) to build robustness to multiple field conditions. A Generative Adversarial Network (GAN)-based augmentation method was imposed to balance class distributions and better generalize the model. Further, a self-supervised contrastive pre-training method helps to learn more features from limited annotated data. Experimental results yield superior results with 99.33% accuracy, precision, recall, and F1-score on multi-benchmark datasets. The proposed model architecture enables local, global, and relational feature representations and offers high interpretability and adaptability. Practically, the framework allows real-time, efficient deployment of edge devices for automated weed detecting, reducing over-reliance on herbicides and providing scalable, sustainable precision-farming options.", "date": "2025-11-20", "categories": []}
{"id": "2511.15535", "title": "A Hybrid CNN-ViT-GNN Framework with GAN-Based Augmentation for Intelligent Weed Detection in Precision Agriculture", "abstract": "The task of weed detection is an essential element of precision agriculture since accurate species identification allows a farmer to selectively apply herbicides and fits into sustainable agriculture crop management. This paper proposes a hybrid deep learning framework recipe for weed detection that utilizes Convolutional Neural Networks (CNNs), Vision Transformers (ViTs), and Graph Neural Networks (GNNs) to build robustness to multiple field conditions. A Generative Adversarial Network (GAN)-based augmentation method was imposed to balance class distributions and better generalize the model. Further, a self-supervised contrastive pre-training method helps to learn more features from limited annotated data. Experimental results yield superior results with 99.33% accuracy, precision, recall, and F1-score on multi-benchmark datasets. The proposed model architecture enables local, global, and relational feature representations and offers high interpretability and adaptability. Practically, the framework allows real-time, efficient deployment to edge devices for automated weed detecting, reducing over-reliance on herbicides and providing scalable, sustainable precision-farming options.", "date": "2025-11-19", "categories": []}
{"id": "2507.10056", "title": "Lightweight Model for Poultry Disease Detection from Fecal Images Using Multi-Color Space Feature Optimization and Machine Learning", "abstract": "Poultry farming is a vital component of the global food supply chain, yet it remains highly vulnerable to infectious diseases such as coccidiosis, salmonellosis, and Newcastle disease. This study proposes a lightweight machine learning-based approach to detect these diseases by analyzing poultry fecal images. We utilize multi-color space feature extraction (RGB, HSV, LAB) and explore a wide range of color, texture, and shape-based descriptors, including color histograms, local binary patterns (LBP), wavelet transforms, and edge detectors. Through a systematic ablation study and dimensionality reduction using PCA and XGBoost feature selection, we identify a compact global feature set that balances accuracy and computational efficiency. An artificial neural network (ANN) classifier trained on these features achieved 95.85% accuracy while requiring no GPU and only 638 seconds of execution time in Google Colab. Compared to deep learning models such as Xception and MobileNetV3, our proposed model offers comparable accuracy with drastically lower resource usage. This work demonstrates a cost-effective, interpretable, and scalable alternative to deep learning for real-time poultry disease detection in low-resource agricultural settings.", "date": "2025-07-14", "categories": []}
{"id": "2511.12999", "title": "Scalable Vision-Guided Crop Yield Estimation", "abstract": "Precise estimation and uncertainty quantification for average crop yields are critical for agricultural monitoring and decision making. Existing data collection methods, such as crop cuts in randomly sampled fields at harvest time, are relatively time-consuming. Thus, we propose an approach based on prediction-powered inference (PPI) to supplement these crop cuts with less time-consuming field photos. After training a computer vision model to predict the ground truth crop cut yields from the photos, we learn a ``control function\" that recalibrates these predictions with the spatial coordinates of each field. This enables fields with photos but not crop cuts to be leveraged to improve the precision of zone-wide average yield estimates. Our control function is learned by training on a dataset of nearly 20,000 real crop cuts and photos of rice and maize fields in sub-Saharan Africa. To improve precision, we pool training observations across different zones within the same first-level subdivision of each country. Our final PPI-based point estimates of the average yield are provably asymptotically unbiased and cannot increase the asymptotic variance beyond that of the natural baseline estimator -- the sample average of the crop cuts -- as the number of fields grows. We also propose a novel bias-corrected and accelerated (BCa) bootstrap to construct accompanying confidence intervals. Even in zones with as few as 20 fields, the point estimates show significant empirical improvement over the baseline, increasing the effective sample size by as much as 73% for rice and by 12-23% for maize. The confidence intervals are accordingly shorter at minimal cost to empirical finite-sample coverage. This demonstrates the potential for relatively low-cost images to make area-based crop insurance more affordable and thus spur investment into sustainable agricultural practices.", "date": "2025-11-17", "categories": []}
{"id": "2403.18334", "title": "DODA: Adapting Object Detectors to Dynamic Agricultural Environments in Real-Time with Diffusion", "abstract": "Object detection has wide applications in agriculture, but domain shifts of diverse environments limit the broader use of the trained models. Existing domain adaptation methods usually require retraining the model for new domains, which is impractical for agricultural applications due to constantly changing environments. In this paper, we propose DODA ($D$iffusion for $O$bject-detection $D$omain Adaptation in $A$griculture), a diffusion-based framework that can adapt the detector to a new domain in just 2 minutes. DODA incorporates external domain embeddings and an improved layout-to-image approach, allowing it to generate high-quality detection data for new domains without additional training. We demonstrate DODA's effectiveness on the Global Wheat Head Detection dataset, where fine-tuning detectors on DODA-generated data yields significant improvements across multiple domains. DODA provides a simple yet powerful solution for agricultural domain adaptation, reducing the barriers for growers to use detection in personalised environments. The code is available at https://github.com/UTokyo-FieldPhenomics-Lab/DODA.", "date": "2024-03-27", "categories": []}
{"id": "2511.04288", "title": "Vision Foundation Models in Agriculture: Toward Domain-Specific Adaptation for Weed Herbicide Trials Assessment", "abstract": "Herbicide field trials require accurate identification of plant species and assessment of herbicide-induced damage across diverse environments. While general-purpose vision foundation models have shown promising results in complex visual domains, their performance can be limited in agriculture, where fine-grained distinctions between species and damage types are critical.\n  In this work, we adapt a general-purpose vision foundation model to herbicide trial characterization. Trained using a self-supervised learning approach on a large, curated agricultural dataset, the model learns rich and transferable representations optimized for herbicide trials images.\n  Our domain-specific model significantly outperforms the best general-purpose foundation model in both species identification (F1 score improvement from 0.91 to 0.94) and damage classification (from 0.26 to 0.33). Under unseen conditions (new locations and other time), it achieves even greater gains (species identification from 0.56 to 0.66; damage classification from 0.17 to 0.27). In domain-shift scenarios, such as drone imagery, it maintains strong performance (species classification from 0.49 to 0.60).\n  Additionally, we show that domain-specific pretraining enhances segmentation accuracy, particularly in low-annotation regimes. An annotation-efficiency analysis reveals that, under unseen conditions, the domain-specific model achieves 5.4% higher F1 score than the general-purpose model, while using 80% fewer labeled samples.\n  These results demonstrate the generalization capabilities of domain-specific foundation models and their potential to significantly reduce manual annotation efforts, offering a scalable and automated solution for herbicide trial analysis.", "date": "2025-11-06", "categories": []}
{"id": "2511.02207", "title": "Object-Centric 3D Gaussian Splatting for Strawberry Plant Reconstruction and Phenotyping", "abstract": "Strawberries are among the most economically significant fruits in the United States, generating over $2 billion in annual farm-gate sales and accounting for approximately 13% of the total fruit production value. Plant phenotyping plays a vital role in selecting superior cultivars by characterizing plant traits such as morphology, canopy structure, and growth dynamics. However, traditional plant phenotyping methods are time-consuming, labor-intensive, and often destructive. Recently, neural rendering techniques, notably Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS), have emerged as powerful frameworks for high-fidelity 3D reconstruction. By capturing a sequence of multi-view images or videos around a target plant, these methods enable non-destructive reconstruction of complex plant architectures. Despite their promise, most current applications of 3DGS in agricultural domains reconstruct the entire scene, including background elements, which introduces noise, increases computational costs, and complicates downstream trait analysis. To address this limitation, we propose a novel object-centric 3D reconstruction framework incorporating a preprocessing pipeline that leverages the Segment Anything Model v2 (SAM-2) and alpha channel background masking to achieve clean strawberry plant reconstructions. This approach produces more accurate geometric representations while substantially reducing computational time. With a background-free reconstruction, our algorithm can automatically estimate important plant traits, such as plant height and canopy width, using DBSCAN clustering and Principal Component Analysis (PCA). Experimental results show that our method outperforms conventional pipelines in both accuracy and efficiency, offering a scalable and non-destructive solution for strawberry plant phenotyping.", "date": "2025-11-04", "categories": []}
{"id": "2511.00548", "title": "Image-based ground distance detection for crop-residue-covered soil", "abstract": "Conservation agriculture features a soil surface covered with crop residues, which brings benefits of improving soil health and saving water. However, one significant challenge in conservation agriculture lies in precisely controlling the seeding depth on the soil covered with crop residues. This is constrained by the lack of ground distance information, since current distance measurement techniques, like laser, ultrasonic, or mechanical displacement sensors, are incapable of differentiating whether the distance information comes from the residue or the soil. This paper presents an image-based method to get the ground distance information for the crop-residues-covered soil. This method is performed with 3D camera and RGB camera, obtaining depth image and color image at the same time. The color image is used to distinguish the different areas of residues and soil and finally generates a mask image. The mask image is applied to the depth image so that only the soil area depth information can be used to calculate the ground distance, and residue areas can be recognized and excluded from ground distance detection. Experimentation shows that this distance measurement method is feasible for real-time implementation, and the measurement error is within plus or minus 3mm. It can be applied in conservation agriculture machinery for precision depth seeding, as well as other depth-control-demanding applications like transplant or tillage.", "date": "2025-11-01", "categories": []}
{"id": "2511.00777", "title": "A Hybrid YOLOv5-SSD IoT-Based Animal Detection System for Durian Plantation Protection", "abstract": "Durian plantation suffers from animal intrusions that cause crop damage and financial loss. The traditional farming practices prove ineffective due to the unavailability of monitoring without human intervention. The fast growth of machine learning and Internet of Things (IoT) technology has led to new ways to detect animals. However, current systems are limited by dependence on single object detection algorithms, less accessible notification platforms, and limited deterrent mechanisms. This research suggests an IoT-enabled animal detection system for durian crops. The system integrates YOLOv5 and SSD object detection algorithms to improve detection accuracy. The system provides real-time monitoring, with detected intrusions automatically reported to farmers via Telegram notifications for rapid response. An automated sound mechanism (e.g., tiger roar) is triggered once the animal is detected. The YOLO+SSD model achieved accuracy rates of elephant, boar, and monkey at 90%, 85% and 70%, respectively. The system shows the highest accuracy in daytime and decreases at night, regardless of whether the image is still or a video. Overall, this study contributes a comprehensive and practical framework that combines detection, notification, and deterrence, paving the way for future innovations in automated farming solutions.", "date": "2025-11-02", "categories": []}
{"id": "2511.01449", "title": "Privacy Preserving Ordinal-Meta Learning with VLMs for Fine-Grained Fruit Quality Prediction", "abstract": "To effectively manage the wastage of perishable fruits, it is crucial to accurately predict their freshness or shelf life using non-invasive methods that rely on visual data. In this regard, deep learning techniques can offer a viable solution. However, obtaining fine-grained fruit freshness labels from experts is costly, leading to a scarcity of data. Closed proprietary Vision Language Models (VLMs), such as Gemini, have demonstrated strong performance in fruit freshness detection task in both zero-shot and few-shot settings. Nonetheless, food retail organizations are unable to utilize these proprietary models due to concerns related to data privacy, while existing open-source VLMs yield sub-optimal performance for the task. Fine-tuning these open-source models with limited data fails to achieve the performance levels of proprietary models. In this work, we introduce a Model-Agnostic Ordinal Meta-Learning (MAOML) algorithm, designed to train smaller VLMs. This approach utilizes meta-learning to address data sparsity and leverages label ordinality, thereby achieving state-of-the-art performance in the fruit freshness classification task under both zero-shot and few-shot settings. Our method achieves an industry-standard accuracy of 92.71%, averaged across all fruits.\n  Keywords: Fruit Quality Prediction, Vision Language Models, Meta Learning, Ordinal Regression", "date": "2025-11-03", "categories": []}
{"id": "2510.24885", "title": "FruitProm: Probabilistic Maturity Estimation and Detection of Fruits and Vegetables", "abstract": "Maturity estimation of fruits and vegetables is a critical task for agricultural automation, directly impacting yield prediction and robotic harvesting. Current deep learning approaches predominantly treat maturity as a discrete classification problem (e.g., unripe, ripe, overripe). This rigid formulation, however, fundamentally conflicts with the continuous nature of the biological ripening process, leading to information loss and ambiguous class boundaries. In this paper, we challenge this paradigm by reframing maturity estimation as a continuous, probabilistic learning task. We propose a novel architectural modification to the state-of-the-art, real-time object detector, RT-DETRv2, by introducing a dedicated probabilistic head. This head enables the model to predict a continuous distribution over the maturity spectrum for each detected object, simultaneously learning the mean maturity state and its associated uncertainty. This uncertainty measure is crucial for downstream decision-making in robotics, providing a confidence score for tasks like selective harvesting. Our model not only provides a far richer and more biologically plausible representation of plant maturity but also maintains exceptional detection performance, achieving a mean Average Precision (mAP) of 85.6\\% on a challenging, large-scale fruit dataset. We demonstrate through extensive experiments that our probabilistic approach offers more granular and accurate maturity assessments than its classification-based counterparts, paving the way for more intelligent, uncertainty-aware automated systems in modern agriculture", "date": "2025-10-28", "categories": []}
{"id": "2510.24456", "title": "A Critical Study towards the Detection of Parkinsons Disease using ML Technologies", "abstract": "The proposed solution is Deep Learning Technique that will be able classify three types of tea leaves diseases from which two diseases are caused by the pests and one due to pathogens (infectious organisms) and environmental conditions and also show the area damaged by a disease in leaves. Namely Red Rust, Helopeltis and Red spider mite respectively. In this paper we have evaluated two models namely SSD MobileNet V2 and Faster R-CNN ResNet50 V1 for the object detection. The SSD MobileNet V2 gave precision of 0.209 for IOU range of 0.50:0.95 with recall of 0.02 on IOU 0.50:0.95 and final mAP of 20.9%. While Faster R-CNN ResNet50 V1 has precision of 0.252 on IOU range of 0.50:0.95 and recall of 0.044 on IOU of 0.50:0.95 with a mAP of 25%, which is better than SSD. Also used Mask R-CNN for Object Instance Segmentation where we have implemented our custom method to calculate the damaged diseased portion of leaves. Keywords: Tea Leaf Disease, Deep Learning, Red Rust, Helopeltis and Red Spider Mite, SSD MobileNet V2, Faster R-CNN ResNet50 V1 and Mask RCNN.", "date": "2025-10-28", "categories": []}
{"id": "2510.24650", "title": "Advancing site-specific disease and pest management in precision agriculture: From reasoning-driven foundation models to adaptive, feedback-based learning", "abstract": "Site-specific disease management (SSDM) in crops has advanced rapidly through machine and deep learning (ML and DL) for real-time computer vision. Research evolved from handcrafted feature extraction to large-scale automated feature learning. With foundation models (FMs), crop disease datasets are now processed in fundamentally new ways. Unlike traditional neural networks, FMs integrate visual and textual data, interpret symptoms in text, reason about symptom-management relationships, and support interactive QA for growers and educators. Adaptive and imitation learning in robotics further enables field-based disease management. This review screened approx. 40 articles on FM applications for SSDM, focusing on large-language models (LLMs) and vision-language models (VLMs), and discussing their role in adaptive learning (AL), reinforcement learning (RL), and digital twin frameworks for targeted spraying. Key findings: (a) FMs are gaining traction with surging literature in 2023-24; (b) VLMs outpace LLMs, with a 5-10x increase in publications; (c) RL and AL are still nascent for smart spraying; (d) digital twins with RL can simulate targeted spraying virtually; (e) addressing the sim-to-real gap is critical for real-world deployment; (f) human-robot collaboration remains limited, especially in human-in-the-loop approaches where robots detect early symptoms and humans validate uncertain cases; (g) multi-modal FMs with real-time feedback will drive next-gen SSDM. For updates, resources, and contributions, visit, https://github.com/nitin-dominic/AgriPathogenDatabase, to submit papers, code, or datasets.", "date": "2025-10-28", "categories": []}
{"id": "2510.22618", "title": "Cross-Species Transfer Learning in Agricultural AI: Evaluating ZebraPose Adaptation for Dairy Cattle Pose Estimation", "abstract": "Pose estimation serves as a cornerstone of computer vision for understanding animal posture, behavior, and welfare. Yet, agricultural applications remain constrained by the scarcity of large, annotated datasets for livestock, especially dairy cattle. This study evaluates the potential and limitations of cross-species transfer learning by adapting ZebraPose - a vision transformer-based model trained on synthetic zebra imagery - for 27-keypoint detection in dairy cows under real barn conditions. Using three configurations - a custom on-farm dataset (375 images, Sussex, New Brunswick, Canada), a subset of the APT-36K benchmark dataset, and their combination, we systematically assessed model accuracy and generalization across environments. While the combined model achieved promising performance (AP = 0.86, AR = 0.87, PCK 0.5 = 0.869) on in-distribution data, substantial generalization failures occurred when applied to unseen barns and cow populations. These findings expose the synthetic-to-real domain gap as a major obstacle to agricultural AI deployment and emphasize that morphological similarity between species is insufficient for cross-domain transfer. The study provides practical insights into dataset diversity, environmental variability, and computational constraints that influence real-world deployment of livestock monitoring systems. We conclude with a call for agriculture-first AI design, prioritizing farm-level realism, cross-environment robustness, and open benchmark datasets to advance trustworthy and scalable animal-centric technologies.", "date": "2025-10-26", "categories": []}
{"id": "2510.22647", "title": "A Critical Study on Tea Leaf Disease Detection using Deep Learning Techniques", "abstract": "The proposed solution is Deep Learning Technique that will be able classify three types of tea leaves diseases from which two diseases are caused by the pests and one due to pathogens (infectious organisms) and environmental conditions and also show the area damaged by a disease in leaves. Namely Red Rust, Helopeltis and Red spider mite respectively. In this paper we have evaluated two models namely SSD MobileNet V2 and Faster R-CNN ResNet50 V1 for the object detection. The SSD MobileNet V2 gave precision of 0.209 for IOU range of 0.50:0.95 with recall of 0.02 on IOU 0.50:0.95 and final mAP of 20.9%. While Faster R-CNN ResNet50 V1 has precision of 0.252 on IOU range of 0.50:0.95 and recall of 0.044 on IOU of 0.50:0.95 with a mAP of 25%, which is better than SSD. Also used Mask R-CNN for Object Instance Segmentation where we have implemented our custom method to calculate the damaged diseased portion of leaves. Keywords: Tea Leaf Disease, Deep Learning, Red Rust, Helopeltis and Red Spider Mite, SSD MobileNet V2, Faster R-CNN ResNet50 V1 and Mask RCNN.", "date": "2025-10-26", "categories": []}
{"id": "2510.22916", "title": "Estimating Pasture Biomass from Top-View Images: A Dataset for Precision Agriculture", "abstract": "Accurate estimation of pasture biomass is important for decision-making in livestock production systems. Estimates of pasture biomass can be used to manage stocking rates to maximise pasture utilisation, while minimising the risk of overgrazing and promoting overall system health. We present a comprehensive dataset of 1,162 annotated top-view images of pastures collected across 19 locations in Australia. The images were taken across multiple seasons and include a range of temperate pasture species. Each image captures a 70cm * 30cm quadrat and is paired with on-ground measurements including biomass sorted by component (green, dead, and legume fraction), vegetation height, and Normalized Difference Vegetation Index (NDVI) from Active Optical Sensors (AOS). The multidimensional nature of the data, which combines visual, spectral, and structural information, opens up new possibilities for advancing the use of precision grazing management. The dataset is released and hosted in a Kaggle competition that challenges the international Machine Learning community with the task of pasture biomass estimation. The dataset is available on the official Kaggle webpage: https://www.kaggle.com/competitions/csiro-biomass", "date": "2025-10-27", "categories": []}
{"id": "2503.13777", "title": "8-Calves Image dataset", "abstract": "Automated livestock monitoring is crucial for precision farming, but robust computer vision models are hindered by a lack of datasets reflecting real-world group challenges. We introduce the 8-Calves dataset, a challenging benchmark for multi-animal detection, tracking, and identification. It features a one-hour video of eight Holstein Friesian calves in a barn, with frequent occlusions, motion blur, and diverse poses. A semi-automated pipeline using a fine-tuned YOLOv8 detector and ByteTrack, followed by manual correction, provides over 537,000 bounding boxes with temporal identity labels.\n  We benchmark 28 object detectors, showing near-perfect performance on a lenient IoU threshold (mAP50: 95.2-98.9%) but significant divergence on stricter metrics (mAP50:95: 56.5-66.4%), highlighting fine-grained localization challenges. Our identification benchmark across 23 models reveals a trade-off: scaling model size improves classification accuracy but compromises retrieval. Smaller architectures like ConvNextV2 Nano achieve the best balance (73.35% accuracy, 50.82% Top-1 KNN). Pre-training focused on semantic learning (e.g., BEiT) yielded superior transferability. For tracking, leading methods achieve high detection accuracy (MOTA > 0.92) but struggle with identity preservation (IDF1 $\\approx$ 0.27), underscoring a key challenge in occlusion-heavy scenarios.\n  The 8-Calves dataset bridges a gap by providing temporal richness and realistic challenges, serving as a resource for advancing agricultural vision models. The dataset and code are available at https://huggingface.co/datasets/tonyFang04/8-calves.", "date": "2025-03-17", "categories": []}
{"id": "2510.18038", "title": "TriggerNet: A Novel Explainable AI Framework for Red Palm Mite Detection and Multi-Model Comparison and Heuristic-Guided Annotation", "abstract": "The red palm mite infestation has become a serious concern, particularly in regions with extensive palm cultivation, leading to reduced productivity and economic losses. Accurate and early identification of mite-infested plants is critical for effective management. The current study focuses on evaluating and comparing the ML model for classifying the affected plants and detecting the infestation. TriggerNet is a novel interpretable AI framework that integrates Grad-CAM, RISE, FullGrad, and TCAV to generate novel visual explanations for deep learning models in plant classification and disease detection. This study applies TriggerNet to address red palm mite (Raoiella indica) infestation, a major threat to palm cultivation and agricultural productivity. A diverse set of RGB images across 11 plant species, Arecanut, Date Palm, Bird of Paradise, Coconut Palm, Ginger, Citrus Tree, Palm Oil, Orchid, Banana Palm, Avocado Tree, and Cast Iron Plant was utilized for training and evaluation. Advanced deep learning models like CNN, EfficientNet, MobileNet, ViT, ResNet50, and InceptionV3, alongside machine learning classifiers such as Random Forest, SVM, and KNN, were employed for plant classification. For disease classification, all plants were categorized into four classes: Healthy, Yellow Spots, Reddish Bronzing, and Silk Webbing. Snorkel was used to efficiently label these disease classes by leveraging heuristic rules and patterns, reducing manual annotation time and improving dataset reliability.", "date": "2025-10-20", "categories": []}
{"id": "2510.16800", "title": "An RGB-D Image Dataset for Lychee Detection and Maturity Classification for Robotic Harvesting", "abstract": "Lychee is a high-value subtropical fruit. The adoption of vision-based harvesting robots can significantly improve productivity while reduce reliance on labor. High-quality data are essential for developing such harvesting robots. However, there are currently no consistently and comprehensively annotated open-source lychee datasets featuring fruits in natural growing environments. To address this, we constructed a dataset to facilitate lychee detection and maturity classification. Color (RGB) images were acquired under diverse weather conditions, and at different times of the day, across multiple lychee varieties, such as Nuomici, Feizixiao, Heiye, and Huaizhi. The dataset encompasses three different ripeness stages and contains 11,414 images, consisting of 878 raw RGB images, 8,780 augmented RGB images, and 1,756 depth images. The images are annotated with 9,658 pairs of lables for lychee detection and maturity classification. To improve annotation consistency, three individuals independently labeled the data, and their results were then aggregated and verified by a fourth reviewer. Detailed statistical analyses were done to examine the dataset. Finally, we performed experiments using three representative deep learning models to evaluate the dataset. It is publicly available for academic", "date": "2025-10-19", "categories": []}
{"id": "2510.17783", "title": "Botany-Bot: Digital Twin Monitoring of Occluded and Underleaf Plant Structures with Gaussian Splats", "abstract": "Commercial plant phenotyping systems using fixed cameras cannot perceive many plant details due to leaf occlusion. In this paper, we present Botany-Bot, a system for building detailed \"annotated digital twins\" of living plants using two stereo cameras, a digital turntable inside a lightbox, an industrial robot arm, and 3D segmentated Gaussian Splat models. We also present robot algorithms for manipulating leaves to take high-resolution indexable images of occluded details such as stem buds and the underside/topside of leaves. Results from experiments suggest that Botany-Bot can segment leaves with 90.8% accuracy, detect leaves with 86.2% accuracy, lift/push leaves with 77.9% accuracy, and take detailed overside/underside images with 77.3% accuracy. Code, videos, and datasets are available at https://berkeleyautomation.github.io/Botany-Bot/.", "date": "2025-10-20", "categories": []}
{"id": "2510.15757", "title": "Poultry Farm Intelligence: An Integrated Multi-Sensor AI Platform for Enhanced Welfare and Productivity", "abstract": "Poultry farming faces increasing pressure to meet productivity targets while ensuring animal welfare and environmental compliance. Yet many small and medium-sized farms lack affordable, integrated tools for continuous monitoring and decision-making, relying instead on manual, reactive inspections. This paper presents Poultry Farm Intelligence (PoultryFI) - a modular, cost-effective platform that integrates six AI-powered modules: Camera Placement Optimizer, Audio-Visual Monitoring, Analytics & Alerting, Real-Time Egg Counting, Production & Profitability Forecasting, and a Recommendation Module.\n  Camera layouts are first optimized offline using evolutionary algorithms for full poultry house coverage with minimal hardware. The Audio-Visual Monitoring module extracts welfare indicators from synchronized video, audio, and feeding data. Analytics & Alerting produces daily summaries and real-time notifications, while Real-Time Egg Counting uses an edge vision model to automate production tracking. Forecasting models predict egg yield and feed consumption up to 10 days in advance, and the Recommendation Module integrates forecasts with weather data to guide environmental and operational adjustments.\n  This is among the first systems to combine low-cost sensing, edge analytics, and prescriptive AI to continuously monitor flocks, predict production, and optimize performance. Field trials demonstrate 100% egg-count accuracy on Raspberry Pi 5, robust anomaly detection, and reliable short-term forecasting. PoultryFI bridges the gap between isolated pilot tools and scalable, farm-wide intelligence, empowering producers to proactively safeguard welfare and profitability.", "date": "2025-10-17", "categories": []}
{"id": "2507.18376", "title": "A Comprehensive Review of Diffusion Models in Smart Agriculture: Progress, Applications, and Challenges", "abstract": "With the global population increasing and arable land resources becoming increasingly limited, smart and precision agriculture have emerged as essential directions for sustainable agricultural development. Artificial intelligence (AI), particularly deep learning models, has been widely adopted in applications such as crop monitoring, pest detection, and yield prediction. Among recent generative models, diffusion models have demonstrated considerable potential in agricultural image processing, data augmentation, and remote sensing analysis. Compared to traditional generative adversarial networks (GANs), diffusion models exhibit greater training stability and superior image generation quality, effectively addressing challenges such as limited annotated datasets and imbalanced sample distributions in agricultural scenarios. This paper reviews recent advancements in the application of diffusion models within agriculture, focusing on their roles in crop disease and pest detection, remote sensing image enhancement, crop growth prediction, and agricultural resource management. Diffusion models have been found useful in improving tasks like image generation, denoising, and data augmentation in agriculture, especially when environmental noise or variability is present. While their high computational requirements and limited generalizability across domains remain concerns, the approach is gradually proving effective in real-world applications such as precision crop monitoring. As research progresses, these models may help support sustainable agriculture and address emerging challenges in food systems.", "date": "2025-07-24", "categories": []}
{"id": "2510.11576", "title": "Benchmarking foundation models for hyperspectral image classification: Application to cereal crop type mapping", "abstract": "Foundation models are transforming Earth observation, but their potential for hyperspectral crop mapping remains underexplored. This study benchmarks three foundation models for cereal crop mapping using hyperspectral imagery: HyperSigma, DOFA, and Vision Transformers pre-trained on the SpectralEarth dataset (a large multitemporal hyperspectral archive). Models were fine-tuned on manually labeled data from a training region and evaluated on an independent test region. Performance was measured with overall accuracy (OA), average accuracy (AA), and F1-score. HyperSigma achieved an OA of 34.5% (+/- 1.8%), DOFA reached 62.6% (+/- 3.5%), and the SpectralEarth model achieved an OA of 93.5% (+/- 0.8%). A compact SpectralEarth variant trained from scratch achieved 91%, highlighting the importance of model architecture for strong generalization across geographic regions and sensor platforms. These results provide a systematic evaluation of foundation models for operational hyperspectral crop mapping and outline directions for future model development.", "date": "2025-10-13", "categories": []}
{"id": "2510.12579", "title": "Unlocking Zero-Shot Plant Segmentation with Pl@ntNet Intelligence", "abstract": "We present a zero-shot segmentation approach for agricultural imagery that leverages Plantnet, a large-scale plant classification model, in conjunction with its DinoV2 backbone and the Segment Anything Model (SAM). Rather than collecting and annotating new datasets, our method exploits Plantnet's specialized plant representations to identify plant regions and produce coarse segmentation masks. These masks are then refined by SAM to yield detailed segmentations. We evaluate on four publicly available datasets of various complexity in terms of contrast including some where the limited size of the training data and complex field conditions often hinder purely supervised methods. Our results show consistent performance gains when using Plantnet-fine-tuned DinoV2 over the base DinoV2 model, as measured by the Jaccard Index (IoU). These findings highlight the potential of combining foundation models with specialized plant-centric models to alleviate the annotation bottleneck and enable effective segmentation in diverse agricultural scenarios.", "date": "2025-10-14", "categories": []}
{"id": "2510.10141", "title": "YOLOv11-Litchi: Efficient Litchi Fruit Detection based on UAV-Captured Agricultural Imagery in Complex Orchard Environments", "abstract": "Litchi is a high-value fruit, yet traditional manual selection methods are increasingly inadequate for modern production demands. Integrating UAV-based aerial imagery with deep learning offers a promising solution to enhance efficiency and reduce costs. This paper introduces YOLOv11-Litchi, a lightweight and robust detection model specifically designed for UAV-based litchi detection. Built upon the YOLOv11 framework, the proposed model addresses key challenges such as small target size, large model parameters hindering deployment, and frequent target occlusion. To tackle these issues, three major innovations are incorporated: a multi-scale residual module to improve contextual feature extraction across scales, a lightweight feature fusion method to reduce model size and computational costs while maintaining high accuracy, and a litchi occlusion detection head to mitigate occlusion effects by emphasizing target regions and suppressing background interference. Experimental results validate the model's effectiveness. YOLOv11-Litchi achieves a parameter size of 6.35 MB - 32.5% smaller than the YOLOv11 baseline - while improving mAP by 2.5% to 90.1% and F1-Score by 1.4% to 85.5%. Additionally, the model achieves a frame rate of 57.2 FPS, meeting real-time detection requirements. These findings demonstrate the suitability of YOLOv11-Litchi for UAV-based litchi detection in complex orchard environments, showcasing its potential for broader applications in precision agriculture.", "date": "2025-10-11", "categories": []}
{"id": "2510.10573", "title": "Deep semi-supervised approach based on consistency regularization and similarity learning for weeds classification", "abstract": "Weed species classification represents an important step for the development of automated targeting systems that allow the adoption of precision agriculture practices. To reduce costs and yield losses caused by their presence. The identification of weeds is a challenging problem due to their shared similarities with crop plants and the variability related to the differences in terms of their types. Along with the variations in relation to changes in field conditions. Moreover, to fully benefit from deep learning-based methods, large fully annotated datasets are needed. This requires time intensive and laborious process for data labeling, which represents a limitation in agricultural applications. Hence, for the aim of improving the utilization of the unlabeled data, regarding conditions of scarcity in terms of the labeled data available during the learning phase and provide robust and high classification performance. We propose a deep semi-supervised approach, that combines consistency regularization with similarity learning. Through our developed deep auto-encoder architecture, experiments realized on the DeepWeeds dataset and inference in noisy conditions demonstrated the effectiveness and robustness of our method in comparison to state-of-the-art fully supervised deep learning models. Furthermore, we carried out ablation studies for an extended analysis of our proposed joint learning strategy.", "date": "2025-10-12", "categories": []}
{"id": "2510.08901", "title": "Modeling Time-Lapse Trajectories to Characterize Cranberry Growth", "abstract": "Change monitoring is an essential task for cranberry farming as it provides both breeders and growers with the ability to analyze growth, predict yield, and make treatment decisions. However, this task is often done manually, requiring significant time on the part of a cranberry grower or breeder. Deep learning based change monitoring holds promise, despite the caveat of hard-to-interpret high dimensional features and hand-annotations for fine-tuning. To address this gap, we introduce a method for modeling crop growth based on fine-tuning vision transformers (ViTs) using a self-supervised approach that avoids tedious image annotations. We use a two-fold pretext task (time regression and class prediction) to learn a latent space for the time-lapse evolution of plant and fruit appearance. The resulting 2D temporal tracks provide an interpretable time-series model of crop growth that can be used to: 1) predict growth over time and 2) distinguish temporal differences of cranberry varieties. We also provide a novel time-lapse dataset of cranberry fruit featuring eight distinct varieties, observed 52 times over the growing season (span of around four months), annotated with information about fungicide application, yield, and rot. Our approach is general and can be applied to other crops and applications (code and dataset can be found at https://github. com/ronan-39/tlt/).", "date": "2025-10-10", "categories": []}
{"id": "2510.07580", "title": "MaizeStandCounting (MaSC): Automated and Accurate Maize Stand Counting from UAV Imagery Using Image Processing and Deep Learning", "abstract": "Accurate maize stand counts are essential for crop management and research, informing yield prediction, planting density optimization, and early detection of germination issues. Manual counting is labor-intensive, slow, and error-prone, especially across large or variable fields. We present MaizeStandCounting (MaSC), a robust algorithm for automated maize seedling stand counting from RGB imagery captured by low-cost UAVs and processed on affordable hardware. MaSC operates in two modes: (1) mosaic images divided into patches, and (2) raw video frames aligned using homography matrices. Both modes use a lightweight YOLOv9 model trained to detect maize seedlings from V2-V10 growth stages. MaSC distinguishes maize from weeds and other vegetation, then performs row and range segmentation based on the spatial distribution of detections to produce precise row-wise stand counts. Evaluation against in-field manual counts from our 2024 summer nursery showed strong agreement with ground truth (R^2= 0.616 for mosaics, R^2 = 0.906 for raw frames). MaSC processed 83 full-resolution frames in 60.63 s, including inference and post-processing, highlighting its potential for real-time operation. These results demonstrate MaSC's effectiveness as a scalable, low-cost, and accurate tool for automated maize stand counting in both research and production environments.", "date": "2025-10-08", "categories": []}
{"id": "2510.07008", "title": "Bayesian Modelling of Multi-Year Crop Type Classification Using Deep Neural Networks and Hidden Markov Models", "abstract": "The temporal consistency of yearly land-cover maps is of great importance to model the evolution and change of the land cover over the years. In this paper, we focus the attention on a novel approach to classification of yearly satellite image time series (SITS) that combines deep learning with Bayesian modelling, using Hidden Markov Models (HMMs) integrated with Transformer Encoder (TE) based DNNs. The proposed approach aims to capture both i) intricate temporal correlations in yearly SITS and ii) specific patterns in multiyear crop type sequences. It leverages the cascade classification of an HMM layer built on top of the TE, discerning consistent yearly crop-type sequences. Validation on a multiyear crop type classification dataset spanning 47 crop types and six years of Sentinel-2 acquisitions demonstrates the importance of modelling temporal consistency in the predicted labels. HMMs enhance the overall performance and F1 scores, emphasising the effectiveness of the proposed approach.", "date": "2025-10-08", "categories": []}
{"id": "2510.05668", "title": "Development and Validation of a Low-Cost Imaging System for Seedling Germination Kinetics through Time-Cumulative Analysis", "abstract": "The study investigates the effects of R. solani inoculation on the germination and early development of Lactuca sativa L. seeds using a low-cost, image-based monitoring system. Multiple cameras were deployed to continuously capture images of the germination process in both infected and control groups. The objective was to assess the impact of the pathogen by analyzing germination dynamics and growth over time. To achieve this, a novel image analysis pipeline was developed. The algorithm integrates both morphological and spatial features to identify and quantify individual seedlings, even under complex conditions where traditional image analyses fails. A key innovation of the method lies in its temporal integration: each analysis step considers not only the current status but also their developmental across prior time points. This approach enables robust discrimination of individual seedlings, especially when overlapping leaves significantly hinder object separation. The method demonstrated high accuracy in seedling counting and vigor assessment, even in challenging scenarios characterized by dense and intertwined growth. Results confirm that R. solani infection significantly reduces germination rates and early seedling vigor. The study also validates the feasibility of combining low-cost imaging hardware with advanced computational tools to obtain phenotyping data in a non-destructive and scalable manner. The temporal integration enabled accurate quantification of germinated seeds and precise determination of seedling emergence timing. This approach proved particularly effective in later stages of the experiment, where conventional segmentation techniques failed due to overlapping or intertwined seedlings, making accurate counting. The method achieved a coefficient of determination of 0.98 and a root mean square error (RMSE) of 1.12, demonstrating its robustness and reliability.", "date": "2025-10-07", "categories": []}
{"id": "2412.10525", "title": "RowDetr: End-to-End Crop Row Detection Using Polynomials", "abstract": "Crop row detection enables autonomous robots to navigate in gps denied environments. Vision based strategies often struggle in the environments due to gaps, curved crop rows and require post-processing steps. Furthermore, labeling crop rows in under the canopy environments accurately is very difficult due to occlusions. This study introduces RowDetr, an efficient end-to-end transformer-based neural network for crop row detection in precision agriculture. RowDetr leverages a lightweight backbone and a hybrid encoder to model straight, curved, or occluded crop rows with high precision. Central to the architecture is a novel polynomial representation that enables direct parameterization of crop rows, eliminating computationally expensive post-processing. Key innovations include a PolySampler module and multi-scale deformable attention, which work together with PolyOptLoss, an energy-based loss function designed to optimize geometric alignment between predicted and the annotated crop rows, while also enhancing robustness against labeling noise. RowDetr was evaluated against other state-of-the-art end-to-end crop row detection methods like AgroNav and RolColAttention on a diverse dataset of 6,962 high-resolution images, used for training, validation, and testing across multiple crop types with annotated crop rows. The system demonstrated superior performance, achieved an F1 score up to 0.74 and a lane position deviation as low as 0.405. Furthermore, RowDetr achieves a real-time inference latency of 6.7ms, which was optimized to 3.5ms with INT8 quantization on an NVIDIA Jetson Orin AGX. This work highlighted the critical efficiency of polynomial parameterization, making RowDetr particularly suitable for deployment on edge computing devices in agricultural robotics and autonomous farming equipment. Index terms > Crop Row Detection, Under Canopy Navigation, Transformers, RT-DETR, RT-DETRv2", "date": "2024-12-13", "categories": []}
{"id": "2509.20580", "title": "A Comparative Benchmark of Real-time Detectors for Blueberry Detection towards Precision Orchard Management", "abstract": "Blueberry detection in natural environments remains challenging due to variable lighting, occlusions, and motion blur due to environmental factors and imaging devices. Deep learning-based object detectors promise to address these challenges, but they demand a large-scale, diverse dataset that captures the real-world complexities. Moreover, deploying these models in practical scenarios often requires the right accuracy/speed/memory trade-off in model selection. This study presents a novel comparative benchmark analysis of advanced real-time object detectors, including YOLO (You Only Look Once) (v8-v12) and RT-DETR (Real-Time Detection Transformers) (v1-v2) families, consisting of 36 model variants, evaluated on a newly curated dataset for blueberry detection. This dataset comprises 661 canopy images collected with smartphones during the 2022-2023 seasons, consisting of 85,879 labelled instances (including 36,256 ripe and 49,623 unripe blueberries) across a wide range of lighting conditions, occlusions, and fruit maturity stages. Among the YOLO models, YOLOv12m achieved the best accuracy with a mAP@50 of 93.3%, while RT-DETRv2-X obtained a mAP@50 of 93.6%, the highest among all the RT-DETR variants. The inference time varied with the model scale and complexity, and the mid-sized models appeared to offer a good accuracy-speed balance. To further enhance detection performance, all the models were fine-tuned using Unbiased Mean Teacher-based semi-supervised learning (SSL) on a separate set of 1,035 unlabeled images acquired by a ground-based machine vision platform in 2024. This resulted in accuracy gains ranging from -1.4% to 2.9%, with RT-DETR-v2-X achieving the best mAP@50 of 94.8%. More in-depth research into SSL is needed to better leverage cross-domain unlabeled data. Both the dataset and software programs of this study are made publicly available to support further research.", "date": "2025-09-24", "categories": []}
{"id": "2312.07935", "title": "Comparing YOLOv8 and Mask R-CNN for instance segmentation in complex orchard environments", "abstract": "Instance segmentation is an important image processing operation for agricultural automation, providing precise delineation of individual objects within images and enabling tasks such as selective harvesting and precision pruning. This study compares the one stage YOLOv8 model with the two stage Mask R CNN model for instance segmentation under varying orchard conditions across two datasets. Dataset 1, collected in the dormant season, contains images of apple trees without foliage and was used to train multi object segmentation models delineating branches and trunks. Dataset 2, collected in the early growing season, includes canopy images with green foliage and immature apples and was used to train single object segmentation models delineating fruitlets. Results showed YOLOv8 outperformed Mask R CNN with higher precision and near perfect recall at a confidence threshold of 0.5. For Dataset 1, YOLOv8 achieved precision 0.90 and recall 0.95 compared to 0.81 and 0.81 for Mask R CNN. For Dataset 2, YOLOv8 reached precision 0.93 and recall 0.97 compared to 0.85 and 0.88. Inference times were also lower for YOLOv8, at 10.9 ms and 7.8 ms, versus 15.6 ms and 12.8 ms for Mask R CNN. These findings demonstrate superior accuracy and efficiency of YOLOv8 for real time orchard automation tasks such as robotic harvesting and fruit thinning.", "date": "2023-12-13", "categories": []}
{"id": "2510.02197", "title": "Cross-Breed Pig Identification Using Auricular Vein Pattern Recognition: A Machine Learning Approach for Small-Scale Farming Applications", "abstract": "Accurate livestock identification is a cornerstone of modern farming: it supports health monitoring, breeding programs, and productivity tracking. However, common pig identification methods, such as ear tags and microchips, are often unreliable, costly, target pure breeds, and thus impractical for small-scale farmers. To address this gap, we propose a noninvasive biometric identification approach that leverages uniqueness of the auricular vein patterns. To this end, we have collected 800 ear images from 20 mixed-breed pigs (Landrace cross Pietrain and Duroc cross Pietrain), captured using a standard smartphone and simple back lighting. A multistage computer vision pipeline was developed to enhance vein visibility, extract structural and spatial features, and generate biometric signatures. These features were then classified using machine learning models. Support Vector Machines (SVM) achieved the highest accuracy: correctly identifying pigs with 98.12% precision across mixed-breed populations. The entire process from image processing to classification was completed in an average of 8.3 seconds, demonstrating feasibility for real-time farm deployment. We believe that by replacing fragile physical identifiers with permanent biological markers, this system provides farmers with a cost-effective and stress-free method of animal identification. More broadly, the findings confirm the practicality of auricular vein biometrics for digitizing livestock management, reinforcing its potential to extend the benefits of precision farming to resource-constrained agricultural communities.", "date": "2025-10-02", "categories": []}
{"id": "2502.00205", "title": "EcoWeedNet: A Lightweight and Automated Weed Detection Method for Sustainable Next-Generation Agricultural Consumer Electronics", "abstract": "Sustainable agriculture plays a crucial role in ensuring world food security for consumers. A critical challenge faced by sustainable precision agriculture is weed growth, as weeds compete for essential resources with crops, such as water, soil nutrients, and sunlight, which notably affect crop yields. The adoption of automated computer vision technologies and ground agricultural consumer electronic vehicles in precision agriculture offers sustainable, low-carbon solutions. However, prior works suffer from issues such as low accuracy and precision, as well as high computational expense. This work proposes EcoWeedNet, a novel model that enhances weed detection performance without introducing significant computational complexity, aligning with the goals of low-carbon agricultural practices. The effectiveness of the proposed model is demonstrated through comprehensive experiments on the CottonWeedDet12 benchmark dataset, which reflects real-world scenarios. EcoWeedNet achieves performance comparable to that of large models (mAP@0.5 = 95.2%), yet with significantly fewer parameters (approximately 4.21% of the parameters of YOLOv4), lower computational complexity and better computational efficiency 6.59% of the GFLOPs of YOLOv4). These key findings indicate EcoWeedNet's deployability on low-power consumer hardware, lower energy consumption, and hence reduced carbon footprint, thereby emphasizing the application prospects of EcoWeedNet in next-generation sustainable agriculture. These findings provide the way forward for increased application of environmentally-friendly agricultural consumer technologies.", "date": "2025-01-31", "categories": []}
{"id": "2509.25805", "title": "Adapting SAM with Dynamic Similarity Graphs for Few-Shot Parameter-Efficient Small Dense Object Detection: A Case Study of Chickpea Pods in Field Conditions", "abstract": "Parameter-Efficient Fine-Tuning (PEFT) of foundation models for agricultural computer vision tasks remains challenging due to limited training data and complex field conditions. This study introduces a Dynamic Similarity-based Graph Adaptation (DSGA) module to adapt the Segment Anything Model (SAM) under extreme data constraints for precise foreground and instance segmentation of small dense objects in complex agricultural environments. Through dynamic similarity graph construction with a learnable polynomial decay-initialized weight ranking mechanism and adaptive local feature aggregation, DSGA establishes robust spatial and dynamic similarity representation with only 4.00M trainable parameters, which is 4.26% of the original SAM. Integrating this graph-based feature adaptation with Low-Rank Adaptation (LoRA) creates a complementary optimization framework that effectively captures both local and global dependencies in image embeddings while preserving model stability and parameter efficiency. Experimental results on a challenging chickpea pod dataset demonstrated that DSGA with LoRA achieved superior performance across multiple metrics evaluated under 2, 4, 8 and 10 shots, with progressive performance gains as shot count increased. Quantitative metrics showed a 17.31% improvement in Structure-measure and a 62.36% gain in adaptive F-measure compared to the baseline SAM fine-tuning. Comprehensive ablation studies and visualization analyses through Grad-CAM and t-SNE validated the framework's effectiveness in feature discrimination. The proposed adaptation demonstrated practical utility for automated agricultural monitoring applications, achieving accurate pod-counting with an adjusted R-squared of 0.8987 for images with 10 to 120 pods under challenging field conditions.", "date": "2025-09-30", "categories": []}
{"id": "2509.23891", "title": "LifeCLEF Plant Identification Task 2015", "abstract": "The LifeCLEF plant identification challenge aims at evaluating plant identification methods and systems at a very large scale, close to the conditions of a real-world biodiversity monitoring scenario. The 2015 evaluation was actually conducted on a set of more than 100K images illustrating 1000 plant species living in West Europe. The main originality of this dataset is that it was built through a large-scale participatory sensing plateform initiated in 2011 and which now involves tens of thousands of contributors. This overview presents more precisely the resources and assessments of the challenge, summarizes the approaches and systems employed by the participating research groups, and provides an analysis of the main outcomes.", "date": "2025-09-28", "categories": []}
{"id": "2509.24329", "title": "TP-MVCC: Tri-plane Multi-view Fusion Model for Silkie Chicken Counting", "abstract": "Accurate animal counting is essential for smart farming but remains difficult in crowded scenes due to occlusions and limited camera views. To address this, we propose a tri-plane-based multi-view chicken counting model (TP-MVCC), which leverages geometric projection and tri-plane fusion to integrate features from multiple cameras onto a unified ground plane. The framework extracts single-view features, aligns them via spatial transformation, and decodes a scene-level density map for precise chicken counting. In addition, we construct the first multi-view dataset of silkie chickens under real farming conditions. Experiments show that TP-MVCC significantly outperforms single-view and conventional fusion comparisons, achieving 95.1\\% accuracy and strong robustness in dense, occluded scenarios, demonstrating its practical potential for intelligent agriculture.", "date": "2025-09-29", "categories": []}
{"id": "2509.19990", "title": "SDE-DET: A Precision Network for Shatian Pomelo Detection in Complex Orchard Environments", "abstract": "Pomelo detection is an essential process for their localization, automated robotic harvesting, and maturity analysis. However, detecting Shatian pomelo in complex orchard environments poses significant challenges, including multi-scale issues, obstructions from trunks and leaves, small object detection, etc. To address these issues, this study constructs a custom dataset STP-AgriData and proposes the SDE-DET model for Shatian pomelo detection. SDE-DET first utilizes the Star Block to effectively acquire high-dimensional information without increasing the computational overhead. Furthermore, the presented model adopts Deformable Attention in its backbone, to enhance its ability to detect pomelos under occluded conditions. Finally, multiple Efficient Multi-Scale Attention mechanisms are integrated into our model to reduce the computational overhead and extract deep visual representations, thereby improving the capacity for small object detection. In the experiment, we compared SDE-DET with the Yolo series and other mainstream detection models in Shatian pomelo detection. The presented SDE-DET model achieved scores of 0.883, 0.771, 0.838, 0.497, and 0.823 in Precision, Recall, mAP@0.5, mAP@0.5:0.95 and F1-score, respectively. SDE-DET has achieved state-of-the-art performance on the STP-AgriData dataset. Experiments indicate that the SDE-DET provides a reliable method for Shatian pomelo detection, laying the foundation for the further development of automatic harvest robots.", "date": "2025-09-24", "categories": []}
{"id": "2504.11366", "title": "A Decade of Wheat Mapping for Lebanon", "abstract": "Wheat accounts for approximately 20% of the world's caloric intake, making it a vital component of global food security. Given this importance, mapping wheat fields plays a crucial role in enabling various stakeholders, including policy makers, researchers, and agricultural organizations, to make informed decisions regarding food security, supply chain management, and resource allocation. In this paper, we tackle the problem of accurately mapping wheat fields out of satellite images by introducing an improved pipeline for winter wheat segmentation, as well as presenting a case study on a decade-long analysis of wheat mapping in Lebanon. We integrate a Temporal Spatial Vision Transformer (TSViT) with Parameter-Efficient Fine Tuning (PEFT) and a novel post-processing pipeline based on the Fields of The World (FTW) framework. Our proposed pipeline addresses key challenges encountered in existing approaches, such as the clustering of small agricultural parcels in a single large field. By merging wheat segmentation with precise field boundary extraction, our method produces geometrically coherent and semantically rich maps that enable us to perform in-depth analysis such as tracking crop rotation pattern over years. Extensive evaluations demonstrate improved boundary delineation and field-level precision, establishing the potential of the proposed framework in operational agricultural monitoring and historical trend analysis. By allowing for accurate mapping of wheat fields, this work lays the foundation for a range of critical studies and future advances, including crop monitoring and yield estimation.", "date": "2025-04-15", "categories": []}
{"id": "2509.18697", "title": "Overview of PlantCLEF 2021: cross-domain plant identification", "abstract": "Automated plant identification has improved considerably thanks to recent advances in deep learning and the availability of training data with more and more field photos. However, this profusion of data concerns only a few tens of thousands of species, mainly located in North America and Western Europe, much less in the richest regions in terms of biodiversity such as tropical countries. On the other hand, for several centuries, botanists have systematically collected, catalogued and stored plant specimens in herbaria, especially in tropical regions, and recent efforts by the biodiversity informatics community have made it possible to put millions of digitised records online. The LifeCLEF 2021 plant identification challenge (or \"PlantCLEF 2021\") was designed to assess the extent to which automated identification of flora in data-poor regions can be improved by using herbarium collections. It is based on a dataset of about 1,000 species mainly focused on the Guiana Shield of South America, a region known to have one of the highest plant diversities in the world. The challenge was evaluated as a cross-domain classification task where the training set consisted of several hundred thousand herbarium sheets and a few thousand photos to allow learning a correspondence between the two domains. In addition to the usual metadata (location, date, author, taxonomy), the training data also includes the values of 5 morphological and functional traits for each species. The test set consisted exclusively of photos taken in the field. This article presents the resources and evaluations of the assessment carried out, summarises the approaches and systems used by the participating research groups and provides an analysis of the main results.", "date": "2025-09-23", "categories": []}
{"id": "2509.17044", "title": "AgriDoctor: A Multimodal Intelligent Assistant for Agriculture", "abstract": "Accurate crop disease diagnosis is essential for sustainable agriculture and global food security. Existing methods, which primarily rely on unimodal models such as image-based classifiers and object detectors, are limited in their ability to incorporate domain-specific agricultural knowledge and lack support for interactive, language-based understanding. Recent advances in large language models (LLMs) and large vision-language models (LVLMs) have opened new avenues for multimodal reasoning. However, their performance in agricultural contexts remains limited due to the absence of specialized datasets and insufficient domain adaptation. In this work, we propose AgriDoctor, a modular and extensible multimodal framework designed for intelligent crop disease diagnosis and agricultural knowledge interaction. As a pioneering effort to introduce agent-based multimodal reasoning into the agricultural domain, AgriDoctor offers a novel paradigm for building interactive and domain-adaptive crop health solutions. It integrates five core components: a router, classifier, detector, knowledge retriever and LLMs. To facilitate effective training and evaluation, we construct AgriMM, a comprehensive benchmark comprising 400000 annotated disease images, 831 expert-curated knowledge entries, and 300000 bilingual prompts for intent-driven tool selection. Extensive experiments demonstrate that AgriDoctor, trained on AgriMM, significantly outperforms state-of-the-art LVLMs on fine-grained agricultural tasks, establishing a new paradigm for intelligent and sustainable farming applications.", "date": "2025-09-21", "categories": []}
{"id": "2509.15181", "title": "Maize Seedling Detection Dataset (MSDD): A Curated High-Resolution RGB Dataset for Seedling Maize Detection and Benchmarking with YOLOv9, YOLO11, YOLOv12 and Faster-RCNN", "abstract": "Accurate maize seedling detection is crucial for precision agriculture, yet curated datasets remain scarce. We introduce MSDD, a high-quality aerial image dataset for maize seedling stand counting, with applications in early-season crop monitoring, yield prediction, and in-field management. Stand counting determines how many plants germinated, guiding timely decisions such as replanting or adjusting inputs. Traditional methods are labor-intensive and error-prone, while computer vision enables efficient, accurate detection. MSDD contains three classes-single, double, and triple plants-capturing diverse growth stages, planting setups, soil types, lighting conditions, camera angles, and densities, ensuring robustness for real-world use. Benchmarking shows detection is most reliable during V4-V6 stages and under nadir views. Among tested models, YOLO11 is fastest, while YOLOv9 yields the highest accuracy for single plants. Single plant detection achieves precision up to 0.984 and recall up to 0.873, but detecting doubles and triples remains difficult due to rarity and irregular appearance, often from planting errors. Class imbalance further reduces accuracy in multi-plant detection. Despite these challenges, YOLO11 maintains efficient inference at 35 ms per image, with an additional 120 ms for saving outputs. MSDD establishes a strong foundation for developing models that enhance stand counting, optimize resource allocation, and support real-time decision-making. This dataset marks a step toward automating agricultural monitoring and advancing precision agriculture.", "date": "2025-09-18", "categories": []}
{"id": "2509.12274", "title": "Developing an aeroponic smart experimental greenhouse for controlling irrigation and plant disease detection using deep learning and IoT", "abstract": "Controlling environmental conditions and monitoring plant status in greenhouses is critical to promptly making appropriate management decisions aimed at promoting crop production. The primary objective of this research study was to develop and test a smart aeroponic greenhouse on an experimental scale where the status of Geranium plant and environmental conditions are continuously monitored through the integration of the internet of things (IoT) and artificial intelligence (AI). An IoT-based platform was developed to control the environmental conditions of plants more efficiently and provide insights to users to make informed management decisions. In addition, we developed an AI-based disease detection framework using VGG-19, InceptionResNetV2, and InceptionV3 algorithms to analyze the images captured periodically after an intentional inoculation. The performance of the AI framework was compared with an expert's evaluation of disease status. Preliminary results showed that the IoT system implemented in the greenhouse environment is able to publish data such as temperature, humidity, water flow, and volume of charge tanks online continuously to users and adjust the controlled parameters to provide an optimal growth environment for the plants. Furthermore, the results of the AI framework demonstrate that the VGG-19 algorithm was able to identify drought stress and rust leaves from healthy leaves with the highest accuracy, 92% among the other algorithms.", "date": "2025-09-14", "categories": []}
{"id": "2509.12442", "title": "Cott-ADNet: Lightweight Real-Time Cotton Boll and Flower Detection Under Field Conditions", "abstract": "Cotton is one of the most important natural fiber crops worldwide, yet harvesting remains limited by labor-intensive manual picking, low efficiency, and yield losses from missing the optimal harvest window. Accurate recognition of cotton bolls and their maturity is therefore essential for automation, yield estimation, and breeding research. We propose Cott-ADNet, a lightweight real-time detector tailored to cotton boll and flower recognition under complex field conditions. Building on YOLOv11n, Cott-ADNet enhances spatial representation and robustness through improved convolutional designs, while introducing two new modules: a NeLU-enhanced Global Attention Mechanism to better capture weak and low-contrast features, and a Dilated Receptive Field SPPF to expand receptive fields for more effective multi-scale context modeling at low computational cost. We curate a labeled dataset of 4,966 images, and release an external validation set of 1,216 field images to support future research. Experiments show that Cott-ADNet achieves 91.5% Precision, 89.8% Recall, 93.3% mAP50, 71.3% mAP, and 90.6% F1-Score with only 7.5 GFLOPs, maintaining stable performance under multi-scale and rotational variations. These results demonstrate Cott-ADNet as an accurate and efficient solution for in-field deployment, and thus provide a reliable basis for automated cotton harvesting and high-throughput phenotypic analysis. Code and dataset is available at https://github.com/SweefongWong/Cott-ADNet.", "date": "2025-09-15", "categories": []}
{"id": "2509.12511", "title": "Axis-Aligned 3D Stalk Diameter Estimation from RGB-D Imagery", "abstract": "Accurate, high-throughput phenotyping is a critical component of modern crop breeding programs, especially for improving traits such as mechanical stability, biomass production, and disease resistance. Stalk diameter is a key structural trait, but traditional measurement methods are labor-intensive, error-prone, and unsuitable for scalable phenotyping. In this paper, we present a geometry-aware computer vision pipeline for estimating stalk diameter from RGB-D imagery. Our method integrates deep learning-based instance segmentation, 3D point cloud reconstruction, and axis-aligned slicing via Principal Component Analysis (PCA) to perform robust diameter estimation. By mitigating the effects of curvature, occlusion, and image noise, this approach offers a scalable and reliable solution to support high-throughput phenotyping in breeding and agronomic research.", "date": "2025-09-15", "categories": []}
{"id": "2509.09961", "title": "Augment to Segment: Tackling Pixel-Level Imbalance in Wheat Disease and Pest Segmentation", "abstract": "Accurate segmentation of foliar diseases and insect damage in wheat is crucial for effective crop management and disease control. However, the insect damage typically occupies only a tiny fraction of annotated pixels. This extreme pixel-level imbalance poses a significant challenge to the segmentation performance, which can result in overfitting to common classes and insufficient learning of rare classes, thereby impairing overall performance. In this paper, we propose a Random Projected Copy-and-Paste (RPCP) augmentation technique to address the pixel imbalance problem. Specifically, we extract rare insect-damage patches from annotated training images and apply random geometric transformations to simulate variations. The transformed patches are then pasted in appropriate regions while avoiding overlaps with lesions or existing damaged regions. In addition, we apply a random projection filter to the pasted regions, refining local features and ensuring a natural blend with the new background. Experiments show that our method substantially improves segmentation performance on the insect damage class, while maintaining or even slightly enhancing accuracy on other categories. Our results highlight the effectiveness of targeted augmentation in mitigating extreme pixel imbalance, offering a straightforward yet effective solution for agricultural segmentation problems.", "date": "2025-09-12", "categories": []}
{"id": "2509.09972", "title": "Drone-Based Multispectral Imaging and Deep Learning for Timely Detection of Branched Broomrape in Tomato Farms", "abstract": "This study addresses the escalating threat of branched broomrape (Phelipanche ramosa) to California's tomato industry, which supplies over 90 percent of U.S. processing tomatoes. The parasite's largely underground life cycle makes early detection difficult, while conventional chemical controls are costly, environmentally harmful, and often ineffective. To address this, we combined drone-based multispectral imagery with Long Short-Term Memory (LSTM) deep learning networks, using the Synthetic Minority Over-sampling Technique (SMOTE) to handle class imbalance. Research was conducted on a known broomrape-infested tomato farm in Woodland, Yolo County, CA, across five key growth stages determined by growing degree days (GDD). Multispectral images were processed to isolate tomato canopy reflectance. At 897 GDD, broomrape could be detected with 79.09 percent overall accuracy and 70.36 percent recall without integrating later stages. Incorporating sequential growth stages with LSTM improved detection substantially. The best-performing scenario, which integrated all growth stages with SMOTE augmentation, achieved 88.37 percent overall accuracy and 95.37 percent recall. These results demonstrate the strong potential of temporal multispectral analysis and LSTM networks for early broomrape detection. While further real-world data collection is needed for practical deployment, this study shows that UAV-based multispectral sensing coupled with deep learning could provide a powerful precision agriculture tool to reduce losses and improve sustainability in tomato production.", "date": "2025-09-12", "categories": []}
{"id": "2507.20800", "title": "LanternNet: A Hub-and-Spoke System to Seek and Suppress Spotted Lanternfly Populations", "abstract": "The invasive spotted lanternfly (SLF) poses a significant threat to agriculture and ecosystems, causing widespread damage. Current control methods, such as egg scraping, pesticides, and quarantines, prove labor-intensive, environmentally hazardous, and inadequate for long-term SLF suppression. This research introduces LanternNet, a novel autonomous robotic Hub-and-Spoke system designed for scalable detection and suppression of SLF populations. A central, tree-mimicking hub utilizes a YOLOv8 computer vision model for precise SLF identification. Three specialized robotic spokes perform targeted tasks: pest neutralization, environmental monitoring, and navigation/mapping. Field deployment across multiple infested sites over 5 weeks demonstrated LanternNet's efficacy. Quantitative analysis revealed significant reductions (p < 0.01, paired t-tests) in SLF populations and corresponding improvements in tree health indicators across the majority of test sites. Compared to conventional methods, LanternNet offers substantial cost advantages and improved scalability. Furthermore, the system's adaptability for enhanced autonomy and targeting of other invasive species presents significant potential for broader ecological impact. LanternNet demonstrates the transformative potential of integrating robotics and AI for advanced invasive species management and improved environmental outcomes.", "date": "2025-07-28", "categories": []}
{"id": "2509.06367", "title": "MRD-LiNet: A Novel Lightweight Hybrid CNN with Gradient-Guided Unlearning for Improved Drought Stress Identification", "abstract": "Drought stress is a major threat to global crop productivity, making its early and precise detection essential for sustainable agricultural management. Traditional approaches, though useful, are often time-consuming and labor-intensive, which has motivated the adoption of deep learning methods. In recent years, Convolutional Neural Network (CNN) and Vision Transformer architectures have been widely explored for drought stress identification; however, these models generally rely on a large number of trainable parameters, restricting their use in resource-limited and real-time agricultural settings. To address this challenge, we propose a novel lightweight hybrid CNN framework inspired by ResNet, DenseNet, and MobileNet architectures. The framework achieves a remarkable 15-fold reduction in trainable parameters compared to conventional CNN and Vision Transformer models, while maintaining competitive accuracy. In addition, we introduce a machine unlearning mechanism based on a gradient norm-based influence function, which enables targeted removal of specific training data influence, thereby improving model adaptability. The method was evaluated on an aerial image dataset of potato fields with expert-annotated healthy and drought-stressed regions. Experimental results show that our framework achieves high accuracy while substantially lowering computational costs. These findings highlight its potential as a practical, scalable, and adaptive solution for drought stress monitoring in precision agriculture, particularly under resource-constrained conditions.", "date": "2025-09-08", "categories": []}
{"id": "2509.06907", "title": "FoMo4Wheat: Toward reliable crop vision foundation models with globally curated data", "abstract": "Vision-driven field monitoring is central to digital agriculture, yet models built on general-domain pretrained backbones often fail to generalize across tasks, owing to the interaction of fine, variable canopy structures with fluctuating field conditions. We present FoMo4Wheat, one of the first crop-domain vision foundation model pretrained with self-supervision on ImAg4Wheat, the largest and most diverse wheat image dataset to date (2.5 million high-resolution images collected over a decade at 30 global sites, spanning >2,000 genotypes and >500 environmental conditions). This wheat-specific pretraining yields representations that are robust for wheat and transferable to other crops and weeds. Across ten in-field vision tasks at canopy and organ levels, FoMo4Wheat models consistently outperform state-of-the-art models pretrained on general-domain dataset. These results demonstrate the value of crop-specific foundation models for reliable in-field perception and chart a path toward a universal crop foundation model with cross-species and cross-task capabilities. FoMo4Wheat models and the ImAg4Wheat dataset are publicly available online: https://github.com/PheniX-Lab/FoMo4Wheat and https://huggingface.co/PheniX-Lab/FoMo4Wheat. The demonstration website is: https://fomo4wheat.phenix-lab.com/.", "date": "2025-09-08", "categories": []}
{"id": "2506.05972", "title": "Domain Adaptation for Big Data in Agricultural Image Analysis: A Comprehensive Review", "abstract": "With the wide application of computer vision in agriculture, image analysis has become the key to tasks such as crop health monitoring and pest detection. However, the significant domain shifts caused by environmental changes, different crop types, and diverse data acquisition methods seriously hinder the generalization ability of the model in cross-region, cross-season, and complex agricultural scenarios. This paper explores how domain adaptation (DA) techniques can address these challenges to improve cross-domain transferability in agricultural image analysis. DA is considered a promising solution in the case of limited labeled data, insufficient model adaptability, and dynamic changes in the field environment. This paper systematically reviews the latest advances in DA in agricultural images in recent years, focusing on application scenarios such as crop health monitoring, pest and disease detection, and fruit identification, in which DA methods have significantly improved cross-domain performance. We categorize DA methods into shallow learning and deep learning methods, including supervised, semi-supervised and unsupervised strategies, and pay special attention to the adversarial learning-based techniques that perform well in complex scenarios. In addition, this paper also reviews the main public datasets of agricultural images, and evaluates their advantages and limitations in DA research. Overall, this study provides a complete framework and some key insights that can be used as a reference for the research and development of domain adaptation methods in future agricultural vision tasks.", "date": "2025-06-06", "categories": []}
{"id": "2509.01344", "title": "AgroSense: An Integrated Deep Learning System for Crop Recommendation via Soil Image Analysis and Nutrient Profiling", "abstract": "Meeting the increasing global demand for food security and sustainable farming requires intelligent crop recommendation systems that operate in real time. Traditional soil analysis techniques are often slow, labor-intensive, and not suitable for on-field decision-making. To address these limitations, we introduce AgroSense, a deep-learning framework that integrates soil image classification and nutrient profiling to produce accurate and contextually relevant crop recommendations. AgroSense comprises two main components: a Soil Classification Module, which leverages ResNet-18, EfficientNet-B0, and Vision Transformer architectures to categorize soil types from images; and a Crop Recommendation Module, which employs a Multi-Layer Perceptron, XGBoost, LightGBM, and TabNet to analyze structured soil data, including nutrient levels, pH, and rainfall. We curated a multimodal dataset of 10,000 paired samples drawn from publicly available Kaggle repositories, approximately 50,000 soil images across seven classes, and 25,000 nutrient profiles for experimental evaluation. The fused model achieves 98.0% accuracy, with a precision of 97.8%, a recall of 97.7%, and an F1-score of 96.75%, while RMSE and MAE drop to 0.32 and 0.27, respectively. Ablation studies underscore the critical role of multimodal coupling, and statistical validation via t-tests and ANOVA confirms the significance of our improvements. AgroSense offers a practical, scalable solution for real-time decision support in precision agriculture and paves the way for future lightweight multimodal AI systems in resource-constrained environments.", "date": "2025-09-01", "categories": []}
{"id": "2508.17117", "title": "PlantVillageVQA: A Visual Question Answering Dataset for Benchmarking Vision-Language Models in Plant Science", "abstract": "PlantVillageVQA is a large-scale visual question answering (VQA) dataset derived from the widely used PlantVillage image corpus. It was designed to advance the development and evaluation of vision-language models for agricultural decision-making and analysis. The PlantVillageVQA dataset comprises 193,609 high-quality question-answer (QA) pairs grounded over 55,448 images spanning 14 crop species and 38 disease conditions. Questions are organised into 3 levels of cognitive complexity and 9 distinct categories. Each question category was phrased manually following expert guidance and generated via an automated two-stage pipeline: (1) template-based QA synthesis from image metadata and (2) multi-stage linguistic re-engineering. The dataset was iteratively reviewed by domain experts for scientific accuracy and relevancy. The final dataset was evaluated using three state-of-the-art models for quality assessment. Our objective remains to provide a publicly available, standardised and expert-verified database to enhance diagnostic accuracy for plant disease identifications and advance scientific research in the agricultural domain. Our dataset will be open-sourced at https://huggingface.co/datasets/SyedNazmusSakib/PlantVillageVQA.", "date": "2025-08-23", "categories": []}
{"id": "2508.20232", "title": "ATMS-KD: Adaptive Temperature and Mixed Sample Knowledge Distillation for a Lightweight Residual CNN in Agricultural Embedded Systems", "abstract": "This study proposes ATMS-KD (Adaptive Temperature and Mixed-Sample Knowledge Distillation), a novel framework for developing lightweight CNN models suitable for resource-constrained agricultural environments. The framework combines adaptive temperature scheduling with mixed-sample augmentation to transfer knowledge from a MobileNetV3 Large teacher model (5.7\\,M parameters) to lightweight residual CNN students. Three student configurations were evaluated: Compact (1.3\\,M parameters), Standard (2.4\\,M parameters), and Enhanced (3.8\\,M parameters). The dataset used in this study consists of images of \\textit{Rosa damascena} (Damask rose) collected from agricultural fields in the Dades Oasis, southeastern Morocco, providing a realistic benchmark for agricultural computer vision applications under diverse environmental conditions. Experimental evaluation on the Damascena rose maturity classification dataset demonstrated significant improvements over direct training methods. All student models achieved validation accuracies exceeding 96.7\\% with ATMS-KD compared to 95--96\\% with direct training. The framework outperformed eleven established knowledge distillation methods, achieving 97.11\\% accuracy with the compact model -- a 1.60 percentage point improvement over the second-best approach while maintaining the lowest inference latency of 72.19\\,ms. Knowledge retention rates exceeded 99\\% for all configurations, demonstrating effective knowledge transfer regardless of student model capacity.", "date": "2025-08-27", "categories": []}
{"id": "2508.20954", "title": "Olive Tree Satellite Image Segmentation Based On SAM and Multi-Phase Refinement", "abstract": "In the context of proven climate change, maintaining olive biodiversity through early anomaly detection and treatment using remote sensing technology is crucial, offering effective management solutions. This paper presents an innovative approach to olive tree segmentation from satellite images. By leveraging foundational models and advanced segmentation techniques, the study integrates the Segment Anything Model (SAM) to accurately identify and segment olive trees in agricultural plots. The methodology includes SAM segmentation and corrections based on trees alignement in the field and a learanble constraint about the shape and the size. Our approach achieved a 98\\% accuracy rate, significantly surpassing the initial SAM performance of 82\\%.", "date": "2025-08-28", "categories": []}
{"id": "2508.19307", "title": "Advancements in Crop Analysis through Deep Learning and Explainable AI", "abstract": "Rice is a staple food of global importance in terms of trade, nutrition, and economic growth. Among Asian nations such as China, India, Pakistan, Thailand, Vietnam and Indonesia are leading producers of both long and short grain varieties, including basmati, jasmine, arborio, ipsala, and kainat saila. To ensure consumer satisfaction and strengthen national reputations, monitoring rice crops and grain quality is essential. Manual inspection, however, is labour intensive, time consuming and error prone, highlighting the need for automated solutions for quality control and yield improvement. This study proposes an automated approach to classify five rice grain varieties using Convolutional Neural Networks (CNN). A publicly available dataset of 75000 images was used for training and testing. Model evaluation employed accuracy, recall, precision, F1-score, ROC curves, and confusion matrices. Results demonstrated high classification accuracy with minimal misclassifications, confirming the model effectiveness in distinguishing rice varieties. In addition, an accurate diagnostic method for rice leaf diseases such as Brown Spot, Blast, Bacterial Blight, and Tungro was developed. The framework combined explainable artificial intelligence (XAI) with deep learning models including CNN, VGG16, ResNet50, and MobileNetV2. Explainability techniques such as SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations) revealed how specific grain and leaf features influenced predictions, enhancing model transparency and reliability. The findings demonstrate the strong potential of deep learning in agricultural applications, paving the way for robust, interpretable systems that can support automated crop quality inspection and disease diagnosis, ultimately benefiting farmers, consumers, and the agricultural economy.", "date": "2025-08-26", "categories": []}
{"id": "2508.19511", "title": "Weed Detection in Challenging Field Conditions: A Semi-Supervised Framework for Overcoming Shadow Bias and Data Scarcity", "abstract": "The automated management of invasive weeds is critical for sustainable agriculture, yet the performance of deep learning models in real-world fields is often compromised by two factors: challenging environmental conditions and the high cost of data annotation. This study tackles both issues through a diagnostic-driven, semi-supervised framework. Using a unique dataset of approximately 975 labeled and 10,000 unlabeled images of Guinea Grass in sugarcane, we first establish strong supervised baselines for classification (ResNet) and detection (YOLO, RF-DETR), achieving F1 scores up to 0.90 and mAP50 scores exceeding 0.82. Crucially, this foundational analysis, aided by interpretability tools, uncovered a pervasive \"shadow bias,\" where models learned to misidentify shadows as vegetation. This diagnostic insight motivated our primary contribution: a semi-supervised pipeline that leverages unlabeled data to enhance model robustness. By training models on a more diverse set of visual information through pseudo-labeling, this framework not only helps mitigate the shadow bias but also provides a tangible boost in recall, a critical metric for minimizing weed escapes in automated spraying systems. To validate our methodology, we demonstrate its effectiveness in a low-data regime on a public crop-weed benchmark. Our work provides a clear and field-tested framework for developing, diagnosing, and improving robust computer vision systems for the complex realities of precision agriculture.", "date": "2025-08-27", "categories": []}
{"id": "2508.16660", "title": "Optimizing Hyper parameters in CNN for Soil Classification using PSO and Whale Optimization Algorithm", "abstract": "Classifying soil images contributes to better land management, increased agricultural output, and practical solutions for environmental issues. The development of various disciplines, particularly agriculture, civil engineering, and natural resource management, is aided by understanding of soil quality since it helps with risk reduction, performance improvement, and sound decision-making . Artificial intelligence has recently been used in a number of different fields. In this study, an intelligent model was constructed using Convolutional Neural Networks to classify soil kinds, and machine learning algorithms were used to enhance the performance of soil classification . To achieve better implementation and performance of the Convolutional Neural Networks algorithm and obtain valuable results for the process of classifying soil type images, swarm algorithms were employed to obtain the best performance by choosing Hyper parameters for the Convolutional Neural Networks network using the Whale optimization algorithm and the Particle swarm optimization algorithm, and comparing the results of using the two algorithms in the process of multiple classification of soil types. The Accuracy and F1 measures were adopted to test the system, and the results of the proposed work were efficient result", "date": "2025-08-20", "categories": []}
{"id": "2508.17107", "title": "SugarcaneShuffleNet: A Very Fast, Lightweight Convolutional Neural Network for Diagnosis of 15 Sugarcane Leaf Diseases", "abstract": "Despite progress in AI-based plant diagnostics, sugarcane farmers in low-resource regions remain vulnerable to leaf diseases due to the lack of scalable, efficient, and interpretable tools. Many deep learning models fail to generalize under real-world conditions and require substantial computational resources, limiting their use in resource-constrained regions. In this paper, we present SugarcaneLD-BD, a curated dataset for sugarcane leaf-disease classification; SugarcaneShuffleNet, an optimized lightweight model for rapid on-device diagnosis; and SugarcaneAI, a Progressive Web Application for field deployment. SugarcaneLD-BD contains 638 curated images across five classes, including four major sugarcane diseases, collected in Bangladesh under diverse field conditions and verified by expert pathologists. To enhance diversity, we combined SugarcaneLD-BD with two additional datasets, yielding a larger and more representative corpus. Our optimized model, SugarcaneShuffleNet, offers the best trade-off between speed and accuracy for real-time, on-device diagnosis. This 9.26 MB model achieved 98.02% accuracy, an F1-score of 0.98, and an average inference time of 4.14 ms per image. For comparison, we fine-tuned five other lightweight convolutional neural networks: MnasNet, EdgeNeXt, EfficientNet-Lite, MobileNet, and SqueezeNet via transfer learning and Bayesian optimization. MnasNet and EdgeNeXt achieved comparable accuracy to SugarcaneShuffleNet, but required significantly more parameters, memory, and computation, limiting their suitability for low-resource deployment. We integrate SugarcaneShuffleNet into SugarcaneAI, delivering Grad-CAM-based explanations in the field. Together, these contributions offer a diverse benchmark, efficient models for low-resource environments, and a practical tool for sugarcane disease classification. It spans varied lighting, backgrounds and devices used on-farm", "date": "2025-08-23", "categories": []}
{"id": "2508.17305", "title": "First Place Solution to the MLCAS 2025 GWFSS Challenge: The Devil is in the Detail and Minority", "abstract": "In this report, we present our solution during the participation of the MLCAS 2025 GWFSS Challenge. This challenge hosts a semantic segmentation competition specific to wheat plants, which requires to segment three wheat organs including the head, leaf, and stem, and another background class. In 2025, participating a segmentation competition is significantly different from that in previous years where many tricks can play important roles. Nowadays most segmentation tricks have been well integrated into existing codebases such that our naive ViT-Adapter baseline has already achieved sufficiently good performance. Hence, we believe the key to stand out among other competitors is to focus on the problem nature of wheat per se. By probing visualizations, we identify the key -- the stem matters. In contrast to heads and leaves, stems exhibit fine structure and occupy only few pixels, which suffers from fragile predictions and class imbalance. Building on our baseline, we present three technical improvements tailored to stems: i) incorporating a dynamic upsampler SAPA used to enhance detail delineation; ii) leveraging semi-supervised guided distillation with stem-aware sample selection to mine the treasure beneath unlabeled data; and iii) applying a test-time scaling strategy to zoom in and segment twice the image. Despite being simple, the three improvements bring us to the first place of the competition, outperforming the second place by clear margins. Code and models will be released at https://github.com/tiny-smart/gwfss25.", "date": "2025-08-24", "categories": []}
{"id": "2508.14486", "title": "WeedSense: Multi-Task Learning for Weed Segmentation, Height Estimation, and Growth Stage Classification", "abstract": "Weed management represents a critical challenge in agriculture, significantly impacting crop yields and requiring substantial resources for control. Effective weed monitoring and analysis strategies are crucial for implementing sustainable agricultural practices and site-specific management approaches. We introduce WeedSense, a novel multi-task learning architecture for comprehensive weed analysis that jointly performs semantic segmentation, height estimation, and growth stage classification. We present a unique dataset capturing 16 weed species over an 11-week growth cycle with pixel-level annotations, height measurements, and temporal labels. WeedSense leverages a dual-path encoder incorporating Universal Inverted Bottleneck blocks and a Multi-Task Bifurcated Decoder with transformer-based feature fusion to generate multi-scale features and enable simultaneous prediction across multiple tasks. WeedSense outperforms other state-of-the-art models on our comprehensive evaluation. On our multi-task dataset, WeedSense achieves mIoU of 89.78% for segmentation, 1.67cm MAE for height estimation, and 99.99% accuracy for growth stage classification while maintaining real-time inference at 160 FPS. Our multitask approach achieves 3$\\times$ faster inference than sequential single-task execution and uses 32.4% fewer parameters. Please see our project page at weedsense.github.io.", "date": "2025-08-20", "categories": []}
{"id": "2508.14558", "title": "A Comprehensive Review of Agricultural Parcel and Boundary Delineation from Remote Sensing Images: Recent Progress and Future Perspectives", "abstract": "Powered by advances in multiple remote sensing sensors, the production of high spatial resolution images provides great potential to achieve cost-efficient and high-accuracy agricultural inventory and analysis in an automated way. Lots of studies that aim at providing an inventory of the level of each agricultural parcel have generated many methods for Agricultural Parcel and Boundary Delineation (APBD). This review covers APBD methods for detecting and delineating agricultural parcels and systematically reviews the past and present of APBD-related research applied to remote sensing images. With the goal to provide a clear knowledge map of existing APBD efforts, we conduct a comprehensive review of recent APBD papers to build a meta-data analysis, including the algorithm, the study site, the crop type, the sensor type, the evaluation method, etc. We categorize the methods into three classes: (1) traditional image processing methods (including pixel-based, edge-based and region-based); (2) traditional machine learning methods (such as random forest, decision tree); and (3) deep learning-based methods. With deep learning-oriented approaches contributing to a majority, we further discuss deep learning-based methods like semantic segmentation-based, object detection-based and Transformer-based methods. In addition, we discuss five APBD-related issues to further comprehend the APBD domain using remote sensing data, such as multi-sensor data in APBD task, comparisons between single-task learning and multi-task learning in the APBD domain, comparisons among different algorithms and different APBD tasks, etc. Finally, this review proposes some APBD-related applications and a few exciting prospects and potential hot topics in future APBD research. We hope this review help researchers who involved in APBD domain to keep track of its development and tendency.", "date": "2025-08-20", "categories": []}
{"id": "2508.12219", "title": "C2PSA-Enhanced YOLOv11 Architecture: A Novel Approach for Small Target Detection in Cotton Disease Diagnosis", "abstract": "This study presents a deep learning-based optimization of YOLOv11 for cotton disease detection, developing an intelligent monitoring system. Three key challenges are addressed: (1) low precision in early spot detection (35% leakage rate for sub-5mm2 spots), (2) performance degradation in field conditions (25% accuracy drop), and (3) high error rates (34.7%) in multi-disease scenarios. The proposed solutions include: C2PSA module for enhanced small-target feature extraction; Dynamic category weighting to handle sample imbalance; Improved data augmentation via Mosaic-MixUp scaling. Experimental results on a 4,078-image dataset show: mAP50: 0.820 (+8.0% improvement); mAP50-95: 0.705 (+10.5% improvement); Inference speed: 158 FPS. The mobile-deployed system enables real-time disease monitoring and precision treatment in agricultural applications.", "date": "2025-08-17", "categories": []}
{"id": "2405.02218", "title": "Multispectral Fine-Grained Classification of Blackgrass in Wheat and Barley Crops", "abstract": "As the burden of herbicide resistance grows and the environmental costs of excessive herbicide use become clear, new approaches to managing weed populations are needed. This is particularly true for cereal crops, like wheat and barley, that are staple foods and occupy a globally significant share of farmland. Even modest advances in weed management practices across these crops could deliver major benefits for both the environment and food security. Blackgrass is a major grass weed which causes particular problems in cereal crops in north-west Europe, a major cereal production area, because it has high levels of herbicide resistance. Detecting blackgrass is also difficult due to its similarity to cereals. Yet, a systematic review of the literature on weed recognition in wheat and barley, included in this study, highlights that blackgrass - and grass weeds more broadly - have received less research attention compared to certain broadleaf weeds. With the use of machine vision and multispectral imaging, we investigate the effectiveness of state-of-the-art methods to identify blackgrass in wheat and barley crops. As part of this work, we present the Eastern England Blackgrass Dataset, a large dataset with which we evaluate several key aspects of blackgrass weed recognition. Firstly, we determine the performance of different CNN and transformer-based architectures on images from unseen fields. Secondly, we demonstrate the role that different spectral bands have on the performance of weed classification. Lastly, we evaluate the role of dataset size in classification performance for each of the models trialled. All models tested achieved an accuracy greater than 80%. Our best model achieved 89.6% and that only half the training data was required to achieve this performance. Our dataset is available at: https://lcas.lincoln.ac.uk/wp/research/data-sets-software/eastern-england-blackgrass-dataset .", "date": "2024-05-03", "categories": []}
{"id": "2508.10156", "title": "Improving watermelon (Citrullus lanatus) disease classification with generative artificial intelligence (GenAI)-based synthetic and real-field images via a custom EfficientNetV2-L model", "abstract": "The current advancements in generative artificial intelligence (GenAI) models have paved the way for new possibilities for generating high-resolution synthetic images, thereby offering a promising alternative to traditional image acquisition for training computer vision models in agriculture. In the context of crop disease diagnosis, GenAI models are being used to create synthetic images of various diseases, potentially facilitating model creation and reducing the dependency on resource-intensive in-field data collection. However, limited research has been conducted on evaluating the effectiveness of integrating real with synthetic images to improve disease classification performance. Therefore, this study aims to investigate whether combining a limited number of real images with synthetic images can enhance the prediction accuracy of an EfficientNetV2-L model for classifying watermelon \\textit{(Citrullus lanatus)} diseases. The training dataset was divided into five treatments: H0 (only real images), H1 (only synthetic images), H2 (1:1 real-to-synthetic), H3 (1:10 real-to-synthetic), and H4 (H3 + random images to improve variability and model generalization). All treatments were trained using a custom EfficientNetV2-L architecture with enhanced fine-tuning and transfer learning techniques. Models trained on H2, H3, and H4 treatments demonstrated high precision, recall, and F1-score metrics. Additionally, the weighted F1-score increased from 0.65 (on H0) to 1.00 (on H3-H4) signifying that the addition of a small number of real images with a considerable volume of synthetic images improved model performance and generalizability. Overall, this validates the findings that synthetic images alone cannot adequately substitute for real images; instead, both must be used in a hybrid manner to maximize model performance for crop disease classification.", "date": "2025-08-13", "categories": []}
{"id": "2508.10817", "title": "Mobile-Friendly Deep Learning for Plant Disease Detection: A Lightweight CNN Benchmark Across 101 Classes of 33 Crops", "abstract": "Plant diseases are a major threat to food security globally. It is important to develop early detection systems which can accurately detect. The advancement in computer vision techniques has the potential to solve this challenge. We have developed a mobile-friendly solution which can accurately classify 101 plant diseases across 33 crops. We built a comprehensive dataset by combining different datasets, Plant Doc, PlantVillage, and PlantWild, all of which are for the same purpose. We evaluated performance across several lightweight architectures - MobileNetV2, MobileNetV3, MobileNetV3-Large, and EfficientNet-B0, B1 - specifically chosen for their efficiency on resource-constrained devices. The results were promising, with EfficientNet-B1 delivering our best performance at 94.7% classification accuracy. This architecture struck an optimal balance between accuracy and computational efficiency, making it well-suited for real-world deployment on mobile devices.", "date": "2025-08-14", "categories": []}
{"id": "2508.08317", "title": "Evaluation of State-of-the-Art Deep Learning Techniques for Plant Disease and Pest Detection", "abstract": "Addressing plant diseases and pests is critical for enhancing crop production and preventing economic losses. Recent advances in artificial intelligence (AI), machine learning (ML), and deep learning (DL) have significantly improved the precision and efficiency of detection methods, surpassing the limitations of manual identification. This study reviews modern computer-based techniques for detecting plant diseases and pests from images, including recent AI developments. The methodologies are organized into five categories: hyperspectral imaging, non-visualization techniques, visualization approaches, modified deep learning architectures, and transformer models. This structured taxonomy provides researchers with detailed, actionable insights for selecting advanced state-of-the-art detection methods. A comprehensive survey of recent work and comparative studies demonstrates the consistent superiority of modern AI-based approaches, which often outperform older image analysis methods in speed and accuracy. In particular, vision transformers such as the Hierarchical Vision Transformer (HvT) have shown accuracy exceeding 99.3% in plant disease detection, outperforming architectures like MobileNetV3. The study concludes by discussing system design challenges, proposing solutions, and outlining promising directions for future research.", "date": "2025-08-09", "categories": []}
{"id": "2312.14706", "title": "BonnBeetClouds3D: A Dataset Towards Point Cloud-based Organ-level Phenotyping of Sugar Beet Plants under Field Conditions", "abstract": "Agricultural production is facing severe challenges in the next decades induced by climate change and the need for sustainability, reducing its impact on the environment. Advancements in field management through non-chemical weeding by robots in combination with monitoring of crops by autonomous unmanned aerial vehicles (UAVs) and breeding of novel and more resilient crop varieties are helpful to address these challenges. The analysis of plant traits, called phenotyping, is an essential activity in plant breeding, it however involves a great amount of manual labor. With this paper, we address the problem of automatic fine-grained organ-level geometric analysis needed for precision phenotyping. As the availability of real-world data in this domain is relatively scarce, we propose a novel dataset that was acquired using UAVs capturing high-resolution images of a real breeding trial containing 48 plant varieties and therefore covering great morphological and appearance diversity. This enables the development of approaches for autonomous phenotyping that generalize well to different varieties. Based on overlapping high-resolution images from multiple viewing angles, we compute photogrammetric dense point clouds and provide detailed and accurate point-wise labels for plants, leaves, and salient points as the tip and the base. Additionally, we include measurements of phenotypic traits performed by experts from the German Federal Plant Variety Office on the real plants, allowing the evaluation of new approaches not only on segmentation and keypoint detection but also directly on the downstream tasks. The provided labeled point clouds enable fine-grained plant analysis and support further progress in the development of automatic phenotyping approaches, but also enable further research in surface reconstruction, point cloud completion, and semantic interpretation of point clouds.", "date": "2023-12-22", "categories": []}
{"id": "2508.07306", "title": "DragonFruitQualityNet: A Lightweight Convolutional Neural Network for Real-Time Dragon Fruit Quality Inspection on Mobile Devices", "abstract": "Dragon fruit, renowned for its nutritional benefits and economic value, has experienced rising global demand due to its affordability and local availability. As dragon fruit cultivation expands, efficient pre- and post-harvest quality inspection has become essential for improving agricultural productivity and minimizing post-harvest losses. This study presents DragonFruitQualityNet, a lightweight Convolutional Neural Network (CNN) optimized for real-time quality assessment of dragon fruits on mobile devices. We curated a diverse dataset of 13,789 images, integrating self-collected samples with public datasets (dataset from Mendeley Data), and classified them into four categories: fresh, immature, mature, and defective fruits to ensure robust model training. The proposed model achieves an impressive 93.98% accuracy, outperforming existing methods in fruit quality classification. To facilitate practical adoption, we embedded the model into an intuitive mobile application, enabling farmers and agricultural stakeholders to conduct on-device, real-time quality inspections. This research provides an accurate, efficient, and scalable AI-driven solution for dragon fruit quality control, supporting digital agriculture and empowering smallholder farmers with accessible technology. By bridging the gap between research and real-world application, our work advances post-harvest management and promotes sustainable farming practices.", "date": "2025-08-10", "categories": []}
{"id": "2508.08094", "title": "3D Plant Root Skeleton Detection and Extraction", "abstract": "Plant roots typically exhibit a highly complex and dense architecture, incorporating numerous slender lateral roots and branches, which significantly hinders the precise capture and modeling of the entire root system. Additionally, roots often lack sufficient texture and color information, making it difficult to identify and track root traits using visual methods. Previous research on roots has been largely confined to 2D studies; however, exploring the 3D architecture of roots is crucial in botany. Since roots grow in real 3D space, 3D phenotypic information is more critical for studying genetic traits and their impact on root development. We have introduced a 3D root skeleton extraction method that efficiently derives the 3D architecture of plant roots from a few images. This method includes the detection and matching of lateral roots, triangulation to extract the skeletal structure of lateral roots, and the integration of lateral and primary roots. We developed a highly complex root dataset and tested our method on it. The extracted 3D root skeletons showed considerable similarity to the ground truth, validating the effectiveness of the model. This method can play a significant role in automated breeding robots. Through precise 3D root structure analysis, breeding robots can better identify plant phenotypic traits, especially root structure and growth patterns, helping practitioners select seeds with superior root systems. This automated approach not only improves breeding efficiency but also reduces manual intervention, making the breeding process more intelligent and efficient, thus advancing modern agriculture.", "date": "2025-08-11", "categories": []}
{"id": "2507.01494", "title": "Crop Pest Classification Using Deep Learning Techniques: A Review", "abstract": "Insect pests continue to bring a serious threat to crop yields around the world, and traditional methods for monitoring them are often slow, manual, and difficult to scale. In recent years, deep learning has emerged as a powerful solution, with techniques like convolutional neural networks (CNNs), vision transformers (ViTs), and hybrid models gaining popularity for automating pest detection. This review looks at 37 carefully selected studies published between 2018 and 2025, all focused on AI-based pest classification. The selected research is organized by crop type, pest species, model architecture, dataset usage, and key technical challenges. The early studies relied heavily on CNNs but latest work is shifting toward hybrid and transformer-based models that deliver higher accuracy and better contextual understanding. Still, challenges like imbalanced datasets, difficulty in detecting small pests, limited generalizability, and deployment on edge devices remain significant hurdles. Overall, this review offers a structured overview of the field, highlights useful datasets, and outlines the key challenges and future directions for AI-based pest monitoring systems.", "date": "2025-07-02", "categories": []}
{"id": "2505.12069", "title": "MT-CYP-Net: Multi-Task Network for Pixel-Level Crop Yield Prediction Under Very Few Samples", "abstract": "Accurate and fine-grained crop yield prediction plays a crucial role in advancing global agriculture. However, the accuracy of pixel-level yield estimation based on satellite remote sensing data has been constrained by the scarcity of ground truth data. To address this challenge, we propose a novel approach called the Multi-Task Crop Yield Prediction Network (MT-CYP-Net). This framework introduces an effective multi-task feature-sharing strategy, where features extracted from a shared backbone network are simultaneously utilized by both crop yield prediction decoders and crop classification decoders with the ability to fuse information between them. This design allows MT-CYP-Net to be trained with extremely sparse crop yield point labels and crop type labels, while still generating detailed pixel-level crop yield maps. Concretely, we collected 1,859 yield point labels along with corresponding crop type labels and satellite images from eight farms in Heilongjiang Province, China, in 2023, covering soybean, maize, and rice crops, and constructed a sparse crop yield label dataset. MT-CYP-Net is compared with three classical machine learning and deep learning benchmark methods in this dataset. Experimental results not only indicate the superiority of MT-CYP-Net compared to previous methods on multiple types of crops but also demonstrate the potential of deep networks on precise pixel-level crop yield prediction, especially with limited data labels.", "date": "2025-05-17", "categories": []}
{"id": "2506.01109", "title": "CountingFruit: Language-Guided 3D Fruit Counting with Semantic Gaussian Splatting", "abstract": "Accurate 3D fruit counting in orchards is challenging due to heavy occlusion, semantic ambiguity between fruits and surrounding structures, and the high computational cost of volumetric reconstruction. Existing pipelines often rely on multi-view 2D segmentation and dense volumetric sampling, which lead to accumulated fusion errors and slow inference. We introduce FruitLangGS, a language-guided 3D fruit counting framework that reconstructs orchard-scale scenes using an adaptive-density Gaussian Splatting pipeline with radius-aware pruning and tile-based rasterization, enabling scalable 3D representation. During inference, compressed CLIP-aligned semantic vectors embedded in each Gaussian are filtered via a dual-threshold cosine similarity mechanism, retrieving Gaussians relevant to target prompts while suppressing common distractors (e.g., foliage), without requiring retraining or image-space masks. The selected Gaussians are then sampled into dense point clouds and clustered geometrically to estimate fruit instances, remaining robust under severe occlusion and viewpoint variation. Experiments on nine different orchard-scale datasets demonstrate that FruitLangGS consistently outperforms existing pipelines in instance counting recall, avoiding multi-view segmentation fusion errors and achieving up to 99.7% recall on Pfuji-Size_Orch2018 orchard dataset. Ablation studies further confirm that language-conditioned semantic embedding and dual-threshold prompt filtering are essential for suppressing distractors and improving counting accuracy under heavy occlusion. Beyond fruit counting, the same framework enables prompt-driven 3D semantic retrieval without retraining, highlighting the potential of language-guided 3D perception for scalable agricultural scene understanding.", "date": "2025-06-01", "categories": []}
{"id": "2508.03751", "title": "Modular Transformer Architecture for Precision Agriculture Imaging", "abstract": "This paper addresses the critical need for efficient and accurate weed segmentation from drone video in precision agriculture. A quality-aware modular deep-learning framework is proposed that addresses common image degradation by analyzing quality conditions-such as blur and noise-and routing inputs through specialized pre-processing and transformer models optimized for each degradation type. The system first analyzes drone images for noise and blur using Mean Absolute Deviation and the Laplacian. Data is then dynamically routed to one of three vision transformer models: a baseline for clean images, a modified transformer with Fisher Vector encoding for noise reduction, or another with an unrolled Lucy-Richardson decoder to correct blur. This novel routing strategy allows the system to outperform existing CNN-based methods in both segmentation quality and computational efficiency, demonstrating a significant advancement in deep-learning applications for agriculture.", "date": "2025-08-04", "categories": []}
{"id": "2508.04658", "title": "YOLOv8-Based Deep Learning Model for Automated Poultry Disease Detection and Health Monitoring paper", "abstract": "In the poultry industry, detecting chicken illnesses is essential to avoid financial losses. Conventional techniques depend on manual observation, which is laborious and prone to mistakes. Using YOLO v8 a deep learning model for real-time object recognition. This study suggests an AI based approach, by developing a system that analyzes high resolution chicken photos, YOLO v8 detects signs of illness, such as abnormalities in behavior and appearance. A sizable, annotated dataset has been used to train the algorithm, which provides accurate real-time identification of infected chicken and prompt warnings to farm operators for prompt action. By facilitating early infection identification, eliminating the need for human inspection, and enhancing biosecurity in large-scale farms, this AI technology improves chicken health management. The real-time features of YOLO v8 provide a scalable and effective method for improving farm management techniques.", "date": "2025-08-06", "categories": []}
{"id": "2508.00900", "title": "Sparse 3D Perception for Rose Harvesting Robots: A Two-Stage Approach Bridging Simulation and Real-World Applications", "abstract": "The global demand for medicinal plants, such as Damask roses, has surged with population growth, yet labor-intensive harvesting remains a bottleneck for scalability. To address this, we propose a novel 3D perception pipeline tailored for flower-harvesting robots, focusing on sparse 3D localization of rose centers. Our two-stage algorithm first performs 2D point-based detection on stereo images, followed by depth estimation using a lightweight deep neural network. To overcome the challenge of scarce real-world labeled data, we introduce a photorealistic synthetic dataset generated via Blender, simulating a dynamic rose farm environment with precise 3D annotations. This approach minimizes manual labeling costs while enabling robust model training. We evaluate two depth estimation paradigms: a traditional triangulation-based method and our proposed deep learning framework. Results demonstrate the superiority of our method, achieving an F1 score of 95.6% (synthetic) and 74.4% (real) in 2D detection, with a depth estimation error of 3% at a 2-meter range on synthetic data. The pipeline is optimized for computational efficiency, ensuring compatibility with resource-constrained robotic systems. By bridging the domain gap between synthetic and real-world data, this work advances agricultural automation for specialty crops, offering a scalable solution for precision harvesting.", "date": "2025-07-28", "categories": []}
{"id": "2508.00921", "title": "SmartDate: AI-Driven Precision Sorting and Quality Control in Date Fruits", "abstract": "SmartDate is an AI-powered system for automated sorting and quality control of date fruits. It combines deep learning, genetic algorithms, and reinforcement learning to improve classification accuracy and predict shelf life. The system uses high-resolution imaging and Visible-Near-Infrared (VisNIR) spectral sensors to evaluate key features such as moisture, sugar content, and texture. Reinforcement learning enables real-time adaptation to production conditions, while genetic algorithms optimize model parameters. SmartDate achieved 94.5 percent accuracy, 93.1 percent F1-score, and an AUC-ROC of 0.96. The system reduces waste and ensures that only high-quality dates reach the market, setting a new benchmark in smart agriculture.", "date": "2025-07-30", "categories": []}
{"id": "2508.01752", "title": "Vision transformer-based multi-camera multi-object tracking framework for dairy cow monitoring", "abstract": "Activity and behaviour correlate with dairy cow health and welfare, making continual and accurate monitoring crucial for disease identification and farm productivity. Manual observation and frequent assessments are laborious and inconsistent for activity monitoring. In this study, we developed a unique multi-camera, real-time tracking system for indoor-housed Holstein Friesian dairy cows. This technology uses cutting-edge computer vision techniques, including instance segmentation and tracking algorithms to monitor cow activity seamlessly and accurately. An integrated top-down barn panorama was created by geometrically aligning six camera feeds using homographic transformations. The detection phase used a refined YOLO11-m model trained on an overhead cow dataset, obtaining high accuracy (mAP\\@0.50 = 0.97, F1 = 0.95). SAMURAI, an upgraded Segment Anything Model 2.1, generated pixel-precise cow masks for instance segmentation utilizing zero-shot learning and motion-aware memory. Even with occlusion and fluctuating posture, a motion-aware Linear Kalman filter and IoU-based data association reliably identified cows over time for object tracking. The proposed system significantly outperformed Deep SORT Realtime. Multi-Object Tracking Accuracy (MOTA) was 98.7% and 99.3% in two benchmark video sequences, with IDF1 scores above 99% and near-zero identity switches. This unified multi-camera system can track dairy cows in complex interior surroundings in real time, according to our data. The system reduces redundant detections across overlapping cameras, maintains continuity as cows move between viewpoints, with the aim of improving early sickness prediction through activity quantification and behavioural classification.", "date": "2025-08-03", "categories": []}
{"id": "2508.02409", "title": "Hydra: Accurate Multi-Modal Leaf Wetness Sensing with mm-Wave and Camera Fusion", "abstract": "Leaf Wetness Duration (LWD), the time that water remains on leaf surfaces, is crucial in the development of plant diseases. Existing LWD detection lacks standardized measurement techniques, and variations across different plant characteristics limit its effectiveness. Prior research proposes diverse approaches, but they fail to measure real natural leaves directly and lack resilience in various environmental conditions. This reduces the precision and robustness, revealing a notable practical application and effectiveness gap in real-world agricultural settings. This paper presents Hydra, an innovative approach that integrates millimeter-wave (mm-Wave) radar with camera technology to detect leaf wetness by determining if there is water on the leaf. We can measure the time to determine the LWD based on this detection. Firstly, we design a Convolutional Neural Network (CNN) to selectively fuse multiple mm-Wave depth images with an RGB image to generate multiple feature images. Then, we develop a transformer-based encoder to capture the inherent connection among the multiple feature images to generate a feature map, which is further fed to a classifier for detection. Moreover, we augment the dataset during training to generalize our model. Implemented using a frequency-modulated continuous-wave (FMCW) radar within the 76 to 81 GHz band, Hydra's performance is meticulously evaluated on plants, demonstrating the potential to classify leaf wetness with up to 96% accuracy across varying scenarios. Deploying Hydra in the farm, including rainy, dawn, or poorly light nights, it still achieves an accuracy rate of around 90%.", "date": "2025-08-04", "categories": []}
{"id": "2506.18292", "title": "Three-dimentional reconstruction of complex, dynamic population canopy architecture for crops with a novel point cloud completion model: A case study in Brassica napus rapeseed", "abstract": "Quantitative descriptions of the complete canopy architecture are essential for accurately evaluating crop photosynthesis and yield performance to guide ideotype design. Although various sensing technologies have been developed for three-dimensional (3D) reconstruction of individual plants and canopies, they failed to obtain an accurate description of canopy architectures due to severe occlusion among complex canopy architectures. We proposed an effective method for 3D reconstruction of complex, dynamic population canopy architecture for rapeseed crops with a novel point cloud completion model. A complete point cloud generation framework was developed for automated annotation of the training dataset by distinguishing surface points from occluded points within canopies. The crop population point cloud completion network (CP-PCN) was then designed with a multi-resolution dynamic graph convolutional encoder (MRDG) and a point pyramid decoder (PPD) to predict occluded points. To further enhance feature extraction, a dynamic graph convolutional feature extractor (DGCFE) module was proposed to capture structural variations over the whole rapeseed growth period. The results demonstrated that CP-PCN achieved chamfer distance (CD) values of 3.35 cm -4.51 cm over four growth stages, outperforming the state-of-the-art transformer-based method (PoinTr). Ablation studies confirmed the effectiveness of the MRDG and DGCFE modules. Moreover, the validation experiment demonstrated that the silique efficiency index developed from CP-PCN improved the overall accuracy of rapeseed yield prediction by 11.2% compared to that of using incomplete point clouds. The CP-PCN pipeline has the potential to be extended to other crops, significantly advancing the quantitatively analysis of in-field population canopy architectures.", "date": "2025-06-23", "categories": []}
{"id": "2507.23174", "title": "CNN-based solution for mango classification in agricultural environments", "abstract": "This article exemplifies the design of a fruit detection and classification system using Convolutional\n  Neural Networks (CNN). The goal is to develop a system that automatically assesses fruit quality for\n  farm inventory management. Specifically, a method for mango fruit classification was developed using\n  image processing, ensuring both accuracy and efficiency. Resnet-18 was selected as the preliminary\n  architecture for classification, while a cascade detector was used for detection, balancing execution speed\n  and computational resource consumption. Detection and classification results were displayed through a\n  graphical interface developed in MatLab App Designer, streamlining system interaction. The integration\n  of convolutional neural networks and cascade detectors proffers a reliable solution for fruit classification\n  and detection, with potential applications in agricultural quality control.", "date": "2025-07-31", "categories": []}
{"id": "2507.23487", "title": "Online Estimation of Table-Top Grown Strawberry Mass in Field Conditions with Occlusions", "abstract": "Accurate mass estimation of table-top grown strawberries under field conditions remains challenging due to frequent occlusions and pose variations. This study proposes a vision-based pipeline integrating RGB-D sensing and deep learning to enable non-destructive, real-time and online mass estimation. The method employed YOLOv8-Seg for instance segmentation, Cycle-consistent generative adversarial network (CycleGAN) for occluded region completion, and tilt-angle correction to refine frontal projection area calculations. A polynomial regression model then mapped the geometric features to mass. Experiments demonstrated mean mass estimation errors of 8.11% for isolated strawberries and 10.47% for occluded cases. CycleGAN outperformed large mask inpainting (LaMa) model in occlusion recovery, achieving superior pixel area ratios (PAR) (mean: 0.978 vs. 1.112) and higher intersection over union (IoU) scores (92.3% vs. 47.7% in the [0.9-1] range). This approach addresses critical limitations of traditional methods, offering a robust solution for automated harvesting and yield monitoring with complex occlusion patterns.", "date": "2025-07-31", "categories": []}
{"id": "2507.20506", "title": "An Improved YOLOv8 Approach for Small Target Detection of Rice Spikelet Flowering in Field Environments", "abstract": "Accurately detecting rice flowering time is crucial for timely pollination in hybrid rice seed production. This not only enhances pollination efficiency but also ensures higher yields. However, due to the complexity of field environments and the characteristics of rice spikelets, such as their small size and short flowering period, automated and precise recognition remains challenging. To address this, this study proposes a rice spikelet flowering recognition method based on an improved YOLOv8 object detection model. First, a Bidirectional Feature Pyramid Network (BiFPN) replaces the original PANet structure to enhance feature fusion and improve multi-scale feature utilization. Second, to boost small object detection, a p2 small-object detection head is added, using finer feature mapping to reduce feature loss commonly seen in detecting small targets. Given the lack of publicly available datasets for rice spikelet flowering in field conditions, a high-resolution RGB camera and data augmentation techniques are used to construct a dedicated dataset, providing reliable support for model training and testing. Experimental results show that the improved YOLOv8s-p2 model achieves an mAP@0.5 of 65.9%, precision of 67.6%, recall of 61.5%, and F1-score of 64.41%, representing improvements of 3.10%, 8.40%, 10.80%, and 9.79%, respectively, over the baseline YOLOv8. The model also runs at 69 f/s on the test set, meeting practical application requirements. Overall, the improved YOLOv8s-p2 offers high accuracy and speed, providing an effective solution for automated monitoring in hybrid rice seed production.", "date": "2025-07-28", "categories": []}
{"id": "2507.20531", "title": "Low-Cost Machine Vision System for Sorting Green Lentils (Lens Culinaris) Based on Pneumatic Ejection and Deep Learning", "abstract": "This paper presents the design, development, and evaluation of a dynamic grain classification system for green lentils (Lens Culinaris), which leverages computer vision and pneumatic ejection. The system integrates a YOLOv8-based detection model that identifies and locates grains on a conveyor belt, together with a second YOLOv8-based classification model that categorises grains into six classes: Good, Yellow, Broken, Peeled, Dotted, and Reject. This two-stage YOLOv8 pipeline enables accurate, real-time, multi-class categorisation of lentils, implemented on a low-cost, modular hardware platform. The pneumatic ejection mechanism separates defective grains, while an Arduino-based control system coordinates real-time interaction between the vision system and mechanical components. The system operates effectively at a conveyor speed of 59 mm/s, achieving a grain separation accuracy of 87.2%. Despite a limited processing rate of 8 grams per minute, the prototype demonstrates the potential of machine vision for grain sorting and provides a modular foundation for future enhancements.", "date": "2025-07-28", "categories": []}
{"id": "2407.21652", "title": "Spatial Transformer Network YOLO Model for Agricultural Object Detection", "abstract": "Object detection plays a crucial role in the field of computer vision by autonomously locating and identifying objects of interest. The You Only Look Once (YOLO) model is an effective single-shot detector. However, YOLO faces challenges in cluttered or partially occluded scenes and can struggle with small, low-contrast objects. We propose a new method that integrates spatial transformer networks (STNs) into YOLO to improve performance. The proposed STN-YOLO aims to enhance the model's effectiveness by focusing on important areas of the image and improving the spatial invariance of the model before the detection process. Our proposed method improved object detection performance both qualitatively and quantitatively. We explore the impact of different localization networks within the STN module as well as the robustness of the model across different spatial transformations. We apply the STN-YOLO on benchmark datasets for Agricultural object detection as well as a new dataset from a state-of-the-art plant phenotyping greenhouse facility. Our code and dataset are publicly available.", "date": "2024-07-31", "categories": []}
{"id": "2507.18532", "title": "COT-AD: Cotton Analysis Dataset", "abstract": "This paper presents COT-AD, a comprehensive Dataset designed to enhance cotton crop analysis through computer vision. Comprising over 25,000 images captured throughout the cotton growth cycle, with 5,000 annotated images, COT-AD includes aerial imagery for field-scale detection and segmentation and high-resolution DSLR images documenting key diseases. The annotations cover pest and disease recognition, vegetation, and weed analysis, addressing a critical gap in cotton-specific agricultural datasets. COT-AD supports tasks such as classification, segmentation, image restoration, enhancement, deep generative model-based cotton crop synthesis, and early disease management, advancing data-driven crop management", "date": "2025-07-24", "categories": []}
{"id": "2507.13514", "title": "Sugar-Beet Stress Detection using Satellite Image Time Series", "abstract": "Satellite Image Time Series (SITS) data has proven effective for agricultural tasks due to its rich spectral and temporal nature. In this study, we tackle the task of stress detection in sugar-beet fields using a fully unsupervised approach. We propose a 3D convolutional autoencoder model to extract meaningful features from Sentinel-2 image sequences, combined with acquisition-date-specific temporal encodings to better capture the growth dynamics of sugar-beets. The learned representations are used in a downstream clustering task to separate stressed from healthy fields. The resulting stress detection system can be directly applied to data from different years, offering a practical and accessible tool for stress detection in sugar-beets.", "date": "2025-07-17", "categories": []}
{"id": "2507.09375", "title": "Automated Multi-Class Crop Pathology Classification via Convolutional Neural Networks: A Deep Learning Approach for Real-Time Precision Agriculture", "abstract": "Crop diseases present a significant barrier to agricultural productivity and global food security, especially in large-scale farming where early identification is often delayed or inaccurate. This research introduces a Convolutional Neural Network (CNN)-based image classification system designed to automate the detection and classification of eight common crop diseases using leaf imagery. The methodology involves a complete deep learning pipeline: image acquisition from a large, labeled dataset, preprocessing via resizing, normalization, and augmentation, and model training using TensorFlow with Keras' Sequential API. The CNN architecture comprises three convolutional layers with increasing filter sizes and ReLU activations, followed by max pooling, flattening, and fully connected layers, concluding with a softmax output for multi-class classification. The system achieves high training accuracy (~90%) and demonstrates reliable performance on unseen data, although a validation accuracy of ~60% suggests minor overfitting. Notably, the model integrates a treatment recommendation module, providing actionable guidance by mapping each detected disease to suitable pesticide or fungicide interventions. Furthermore, the solution is deployed on an open-source, mobile-compatible platform, enabling real-time image-based diagnostics for farmers in remote areas. This research contributes a scalable and accessible tool to the field of precision agriculture, reducing reliance on manual inspection and promoting sustainable disease management practices. By merging deep learning with practical agronomic support, this work underscores the potential of CNNs to transform crop health monitoring and enhance food production resilience on a global scale.", "date": "2025-07-12", "categories": []}
{"id": "2507.10499", "title": "National level satellite-based crop field inventories in smallholder landscapes", "abstract": "The design of science-based policies to improve the sustainability of smallholder agriculture is challenged by a limited understanding of fundamental system properties, such as the spatial distribution of active cropland and field size. We integrate very high spatial resolution (1.5 m) Earth observation data and deep transfer learning to derive crop field delineations in complex agricultural systems at the national scale, while maintaining minimum reference data requirements and enhancing transferability. We provide the first national-level dataset of 21 million individual fields for Mozambique (covering ~800,000 km2) for 2023. Our maps separate active cropland from non-agricultural land use with an overall accuracy of 93% and balanced omission and commission errors. Field-level spatial agreement reached median intersection over union (IoU) scores of 0.81, advancing the state-of-the-art in large-area field delineation in complex smallholder systems. The active cropland maps capture fragmented rural regions with low cropland shares not yet identified in global land cover or cropland maps. These regions are mostly located in agricultural frontier regions which host 7-9% of the Mozambican population. Field size in Mozambique is very low overall, with half of the fields being smaller than 0.16 ha, and 83% smaller than 0.5 ha. Mean field size at aggregate spatial resolution (0.05{\\deg}) is 0.32 ha, but it varies strongly across gradients of accessibility, population density, and net forest cover change. This variation reflects a diverse set of actors, ranging from semi-subsistence smallholder farms to medium-scale commercial farming, and large-scale farming operations. Our results highlight that field size is a key indicator relating to socio-economic and environmental outcomes of agriculture (e.g., food production, livelihoods, deforestation, biodiversity), as well as their trade-offs.", "date": "2025-07-14", "categories": []}
{"id": "2502.15286", "title": "Soybean pod and seed counting in both outdoor fields and indoor laboratories using unions of deep neural networks", "abstract": "Automatic counting soybean pods and seeds in outdoor fields allows for rapid yield estimation before harvesting, while indoor laboratory counting offers greater accuracy. Both methods can significantly accelerate the breeding process. However, it remains challenging for accurately counting pods and seeds in outdoor fields, and there are still no accurate enough tools for counting pods and seeds in laboratories. In this study, we developed efficient deep learning models for counting soybean pods and seeds in both outdoor fields and indoor laboratories. For outdoor fields, annotating not only visible seeds but also occluded seeds makes YOLO have the ability to estimate the number of soybean seeds that are occluded. Moreover, we enhanced YOLO architecture by integrating it with HQ-SAM (YOLO-SAM), and domain adaptation techniques (YOLO-DA), to improve model robustness and generalization across soybean images taken in outdoor fields. Testing on soybean images from the outdoor field, we achieved a mean absolute error (MAE) of 6.13 for pod counting and 10.05 for seed counting. For the indoor setting, we utilized Mask-RCNN supplemented with a Swin Transformer module (Mask-RCNN-Swin), models were trained exclusively on synthetic training images generated from a small set of labeled data. This approach resulted in near-perfect accuracy, with an MAE of 1.07 for pod counting and 1.33 for seed counting across actual laboratory images from two distinct studies.", "date": "2025-02-21", "categories": []}
{"id": "2507.05432", "title": "Robotic System with AI for Real Time Weed Detection, Canopy Aware Spraying, and Droplet Pattern Evaluation", "abstract": "Uniform and excessive herbicide application in modern agriculture contributes to increased input costs, environmental pollution, and the emergence of herbicide resistant weeds. To address these challenges, we developed a vision guided, AI-driven variable rate sprayer system capable of detecting weed presence, estimating canopy size, and dynamically adjusting nozzle activation in real time. The system integrates lightweight YOLO11n and YOLO11n-seg deep learning models, deployed on an NVIDIA Jetson Orin Nano for onboard inference, and uses an Arduino Uno-based relay interface to control solenoid actuated nozzles based on canopy segmentation results. Indoor trials were conducted using 15 potted Hibiscus rosa sinensis plants of varying canopy sizes to simulate a range of weed patch scenarios. The YOLO11n model achieved a mean average precision (mAP@50) of 0.98, with a precision of 0.99 and a recall close to 1.0. The YOLO11n-seg segmentation model achieved a mAP@50 of 0.48, precision of 0.55, and recall of 0.52. System performance was validated using water sensitive paper, which showed an average spray coverage of 24.22% in zones where canopy was present. An upward trend in mean spray coverage from 16.22% for small canopies to 21.46% and 21.65% for medium and large canopies, respectively, demonstrated the system's capability to adjust spray output based on canopy size in real time. These results highlight the potential of combining real time deep learning with low-cost embedded hardware for selective herbicide application. Future work will focus on expanding the detection capabilities to include three common weed species in South Dakota: water hemp (Amaranthus tuberculatus), kochia (Bassia scoparia), and foxtail (Setaria spp.), followed by further validation in both indoor and field trials within soybean and corn production systems.", "date": "2025-07-07", "categories": []}
{"id": "2507.06093", "title": "Tile-Based ViT Inference with Visual-Cluster Priors for Zero-Shot Multi-Species Plant Identification", "abstract": "We describe DS@GT's second-place solution to the PlantCLEF 2025 challenge on multi-species plant identification in vegetation quadrat images. Our pipeline combines (i) a fine-tuned Vision Transformer ViTD2PC24All for patch-level inference, (ii) a 4x4 tiling strategy that aligns patch size with the network's 518x518 receptive field, and (iii) domain-prior adaptation through PaCMAP + K-Means visual clustering and geolocation filtering. Tile predictions are aggregated by majority vote and re-weighted with cluster-specific Bayesian priors, yielding a macro-averaged F1 of 0.348 (private leaderboard) while requiring no additional training. All code, configuration files, and reproducibility scripts are publicly available at https://github.com/dsgt-arc/plantclef-2025.", "date": "2025-07-08", "categories": []}
{"id": "2507.03198", "title": "AI-driven Web Application for Early Detection of Sudden Death Syndrome (SDS) in Soybean Leaves Using Hyperspectral Images and Genetic Algorithm", "abstract": "Sudden Death Syndrome (SDS), caused by Fusarium virguliforme, poses a significant threat to soybean production. This study presents an AI-driven web application for early detection of SDS on soybean leaves using hyperspectral imaging, enabling diagnosis prior to visible symptom onset. Leaf samples from healthy and inoculated plants were scanned using a portable hyperspectral imaging system (398-1011 nm), and a Genetic Algorithm was employed to select five informative wavelengths (505.4, 563.7, 712.2, 812.9, and 908.4 nm) critical for discriminating infection status. These selected bands were fed into a lightweight Convolutional Neural Network (CNN) to extract spatial-spectral features, which were subsequently classified using ten classical machine learning models. Ensemble classifiers (Random Forest, AdaBoost), Linear SVM, and Neural Net achieved the highest accuracy (>98%) and minimal error across all folds, as confirmed by confusion matrices and cross-validation metrics. Poor performance by Gaussian Process and QDA highlighted their unsuitability for this dataset. The trained models were deployed within a web application that enables users to upload hyperspectral leaf images, visualize spectral profiles, and receive real-time classification results. This system supports rapid and accessible plant disease diagnostics, contributing to precision agriculture practices. Future work will expand the training dataset to encompass diverse genotypes, field conditions, and disease stages, and will extend the system for multiclass disease classification and broader crop applicability.", "date": "2025-07-03", "categories": []}
{"id": "2507.03219", "title": "Development of an Improved Capsule-Yolo Network for Automatic Tomato Plant Disease Early Detection and Diagnosis", "abstract": "Like many countries, Nigeria is naturally endowed with fertile agricultural soil that supports large-scale tomato production. However, the prevalence of disease causing pathogens poses a significant threat to tomato health, often leading to reduced yields and, in severe cases, the extinction of certain species. These diseases jeopardise both the quality and quantity of tomato harvests, contributing to food insecurity. Fortunately, tomato diseases can often be visually identified through distinct forms, appearances, or textures, typically first visible on leaves and fruits. This study presents an enhanced Capsule-YOLO network architecture designed to automatically segment overlapping and occluded tomato leaf images from complex backgrounds using the YOLO framework. It identifies disease symptoms with impressive performance metrics: 99.31% accuracy, 98.78% recall, and 99.09% precision, and a 98.93% F1-score representing improvements of 2.91%, 1.84%, 5.64%, and 4.12% over existing state-of-the-art methods. Additionally, a user-friendly interface was developed to allow farmers and users to upload images of affected tomato plants and detect early disease symptoms. The system also provides recommendations for appropriate diagnosis and treatment. The effectiveness of this approach promises significant benefits for the agricultural sector by enhancing crop yields and strengthening food security.", "date": "2025-07-03", "categories": []}
{"id": "2507.04366", "title": "Time2Agri: Temporal Pretext Tasks for Agricultural Monitoring", "abstract": "Self Supervised Learning(SSL) has emerged as a prominent paradigm for label-efficient learning, and has been widely utilized by remote sensing foundation models(RSFMs). Recent RSFMs including SatMAE, DoFA, primarily rely on masked autoencoding(MAE), contrastive learning or some combination of them. However, these pretext tasks often overlook the unique temporal characteristics of agricultural landscape, namely nature's cycle. Motivated by this gap, we propose three novel agriculture-specific pretext tasks, namely Time-Difference Prediction(TD), Temporal Frequency Prediction(FP), and Future-Frame Prediction(FF). Comprehensive evaluation on SICKLE dataset shows FF achieves 69.6% IoU on crop mapping and FP reduces yield prediction error to 30.7% MAPE, outperforming all baselines, and TD remains competitive on most tasks. Further, we also scale FF to the national scale of India, achieving 54.2% IoU outperforming all baselines on field boundary delineation on FTW India dataset.", "date": "2025-07-06", "categories": []}
{"id": "2503.07813", "title": "MaizeField3D: A Curated 3D Point Cloud and Procedural Model Dataset of Field-Grown Maize from a Diversity Panel", "abstract": "The development of artificial intelligence (AI) and machine learning (ML) based tools for 3D phenotyping, especially for maize, has been limited due to the lack of large and diverse 3D datasets. 2D image datasets fail to capture essential structural details such as leaf architecture, plant volume, and spatial arrangements that 3D data provide. To address this limitation, we present MaizeField3D (https://baskargroup.github.io/MaizeField3D/), a curated dataset of 3D point clouds of field-grown maize plants from a diverse genetic panel, designed to be AI-ready for advancing agricultural research. Our dataset includes 1,045 high-quality point clouds of field-grown maize collected using a terrestrial laser scanner (TLS). Point clouds of 520 plants from this dataset were segmented and annotated using a graph-based segmentation method to isolate individual leaves and stalks, ensuring consistent labeling across all samples. This labeled data was then used for fitting procedural models that provide a structured parametric representation of the maize plants. The leaves of the maize plants in the procedural models are represented using Non-Uniform Rational B-Spline (NURBS) surfaces that were generated using a two-step optimization process combining gradient-free and gradient-based methods. We conducted rigorous manual quality control on all datasets, correcting errors in segmentation, ensuring accurate leaf ordering, and validating metadata annotations. The dataset also includes metadata detailing plant morphology and quality, alongside multi-resolution subsampled point cloud data (100k, 50k, 10k points), which can be readily used for different downstream computational tasks. MaizeField3D will serve as a comprehensive foundational dataset for AI-driven phenotyping, plant structural analysis, and 3D applications in agricultural research.", "date": "2025-03-10", "categories": []}
{"id": "2507.02517", "title": "Detecting Multiple Diseases in Multiple Crops Using Deep Learning", "abstract": "India, as a predominantly agrarian economy, faces significant challenges in agriculture, including substantial crop losses caused by diseases, pests, and environmental stress. Early detection and accurate identification of diseases across different crops are critical for improving yield and ensuring food security. This paper proposes a deep learning based solution for detecting multiple diseases in multiple crops, aimed to cover India's diverse agricultural landscape. We first create a unified dataset encompassing images of 17 different crops and 34 different diseases from various available repositories. Proposed deep learning model is trained on this dataset and outperforms the state-of-the-art in terms of accuracy and the number of crops, diseases covered. We achieve a significant detection accuracy, i.e., 99 percent for our unified dataset which is 7 percent more when compared to state-of-the-art handling 14 crops and 26 different diseases only. By improving the number of crops and types of diseases that can be detected, proposed solution aims to provide a better product for Indian farmers.", "date": "2025-07-03", "categories": []}
{"id": "2507.01912", "title": "3D Reconstruction and Information Fusion between Dormant and Canopy Seasons in Commercial Orchards Using Deep Learning and Fast GICP", "abstract": "In orchard automation, dense foliage during the canopy season severely occludes tree structures, minimizing visibility to various canopy parts such as trunks and branches, which limits the ability of a machine vision system. However, canopy structure is more open and visible during the dormant season when trees are defoliated. In this work, we present an information fusion framework that integrates multi-seasonal structural data to support robotic and automated crop load management during the entire growing season. The framework combines high-resolution RGB-D imagery from both dormant and canopy periods using YOLOv9-Seg for instance segmentation, Kinect Fusion for 3D reconstruction, and Fast Generalized Iterative Closest Point (Fast GICP) for model alignment. Segmentation outputs from YOLOv9-Seg were used to extract depth-informed masks, which enabled accurate 3D point cloud reconstruction via Kinect Fusion; these reconstructed models from each season were subsequently aligned using Fast GICP to achieve spatially coherent multi-season fusion. The YOLOv9-Seg model, trained on manually annotated images, achieved a mean squared error (MSE) of 0.0047 and segmentation mAP@50 scores up to 0.78 for trunks in dormant season dataset. Kinect Fusion enabled accurate reconstruction of tree geometry, validated with field measurements resulting in root mean square errors (RMSE) of 5.23 mm for trunk diameter, 4.50 mm for branch diameter, and 13.72 mm for branch spacing. Fast GICP achieved precise cross-seasonal registration with a minimum fitness score of 0.00197, allowing integrated, comprehensive tree structure modeling despite heavy occlusions during the growing season. This fused structural representation enables robotic systems to access otherwise obscured architectural information, improving the precision of pruning, thinning, and other automated orchard operations.", "date": "2025-07-02", "categories": []}
{"id": "2412.20521", "title": "Can Robots \"Taste\" Grapes? Estimating SSC with Simple RGB Sensors", "abstract": "In table grape cultivation, harvesting depends on accurately assessing fruit quality. While some characteristics, like color, are visible, others, such as Soluble Solid Content (SSC), or sugar content measured in degrees Brix ({\\deg}Brix), require specific tools. SSC is a key quality factor that correlates with ripeness, but lacks a direct causal relationship with color. Hyperspectral cameras can estimate SSC with high accuracy under controlled laboratory conditions, but their practicality in field environments is limited. This study investigates the potential of simple RGB sensors under uncontrolled lighting to estimate SSC and color, enabling cost-effective, robot-assisted harvesting. Over the 2021 and 2022 summer seasons, we collected grape images with corresponding SSC and color labels to evaluate algorithmic solutions for SSC estimation, specifically testing for cross-seasonal and cross-device robustness. We propose two approaches: a computationally efficient histogram-based method for resource-constrained robots and a Deep Neural Network (DNN) model for more complex applications. Our results demonstrate high performance, with the DNN model achieving a Mean Absolute Error (MAE) as low as $1.05$ {\\deg}Brix on a challenging cross-device test set. The lightweight histogram-based method also proved effective, reaching an MAE of $1.46$ {\\deg}Brix. These results are highly competitive with those from hyperspectral systems, which report errors in the $1.27$--$2.20$ {\\deg}Brix range in similar field applications.", "date": "2024-12-29", "categories": []}
{"id": "2506.22570", "title": "Dual Atrous Separable Convolution for Improving Agricultural Semantic Segmentation", "abstract": "Agricultural image semantic segmentation is a pivotal component of modern agriculture, facilitating accurate visual data analysis to improve crop management, optimize resource utilization, and boost overall productivity. This study proposes an efficient image segmentation method for precision agriculture, focusing on accurately delineating farmland anomalies to support informed decision-making and proactive interventions. A novel Dual Atrous Separable Convolution (DAS Conv) module is integrated within the DeepLabV3-based segmentation framework. The DAS Conv module is meticulously designed to achieve an optimal balance between dilation rates and padding size, thereby enhancing model performance without compromising efficiency. The study also incorporates a strategic skip connection from an optimal stage in the encoder to the decoder to bolster the model's capacity to capture fine-grained spatial features. Despite its lower computational complexity, the proposed model outperforms its baseline and achieves performance comparable to highly complex transformer-based state-of-the-art (SOTA) models on the Agriculture Vision benchmark dataset. It achieves more than 66% improvement in efficiency when considering the trade-off between model complexity and performance, compared to the SOTA model. This study highlights an efficient and effective solution for improving semantic segmentation in remote sensing applications, offering a computationally lightweight model capable of high-quality performance in agricultural imagery.", "date": "2025-06-27", "categories": []}
{"id": "2412.17587", "title": "Improved Cotton Leaf Disease Classification Using Parameter-Efficient Deep Learning Framework", "abstract": "Cotton crops, often called \"white gold,\" face significant production challenges, primarily due to various leaf-affecting diseases. As a major global source of fiber, timely and accurate disease identification is crucial to ensure optimal yields and maintain crop health. While deep learning and machine learning techniques have been explored to address this challenge, there remains a gap in developing lightweight models with fewer parameters which could be computationally effective for agricultural practitioners. To address this, we propose an innovative deep learning framework integrating a subset of trainable layers from MobileNet, transfer learning, data augmentation, a learning rate decay schedule, model checkpoints, and early stopping mechanisms. Our model demonstrates exceptional performance, accurately classifying seven cotton disease types with an overall accuracy of 98.42% and class-wise precision ranging from 96% to 100%. This results in significantly enhanced efficiency, surpassing recent approaches in accuracy and model complexity. The existing models in the literature have yet to attain such high accuracy, even when tested on data sets with fewer disease types. The substantial performance improvement, combined with the lightweight nature of the model, makes it practically suitable for real-world applications in smart farming. By offering a high-performing and efficient solution, our framework can potentially address challenges in cotton cultivation, contributing to sustainable agricultural practices.", "date": "2024-12-23", "categories": []}
{"id": "2506.19939", "title": "Computer Vision based Automated Quantification of Agricultural Sprayers Boom Displacement", "abstract": "Application rate errors when using self-propelled agricultural sprayers for agricultural production remain a concern. Among other factors, spray boom instability is one of the major contributors to application errors. Spray booms' width of 38m, combined with 30 kph driving speeds, varying terrain, and machine dynamics when maneuvering complex field boundaries, make controls of these booms very complex. However, there is no quantitative knowledge on the extent of boom movement to systematically develop a solution that might include boom designs and responsive boom control systems. Therefore, this study was conducted to develop an automated computer vision system to quantify the boom movement of various agricultural sprayers. A computer vision system was developed to track a target on the edge of the sprayer boom in real time. YOLO V7, V8, and V11 neural network models were trained to track the boom's movements in field operations to quantify effective displacement in the vertical and transverse directions. An inclinometer sensor was mounted on the boom to capture boom angles and validate the neural network model output. The results showed that the model could detect the target with more than 90 percent accuracy, and distance estimates of the target on the boom were within 0.026 m of the inclinometer sensor data. This system can quantify the boom movement on the current sprayer and potentially on any other sprayer with minor modifications. The data can be used to make design improvements to make sprayer booms more stable and achieve greater application accuracy.", "date": "2025-06-24", "categories": []}
{"id": "2506.20323", "title": "Comparative Analysis of Deep Learning Models for Crop Disease Detection: A Transfer Learning Approach", "abstract": "This research presents the development of an Artificial Intelligence (AI) - driven crop disease detection system designed to assist farmers in rural areas with limited resources. We aim to compare different deep learning models for a comparative analysis, focusing on their efficacy in transfer learning. By leveraging deep learning models, including EfficientNet, ResNet101, MobileNetV2, and our custom CNN, which achieved a validation accuracy of 95.76%, the system effectively classifies plant diseases. This research demonstrates the potential of transfer learning in reshaping agricultural practices, improving crop health management, and supporting sustainable farming in rural environments.", "date": "2025-06-25", "categories": []}
{"id": "2405.07332", "title": "PotatoGANs: Utilizing Generative Adversarial Networks, Instance Segmentation, and Explainable AI for Enhanced Potato Disease Identification and Classification", "abstract": "Numerous applications have resulted from the automation of agricultural disease segmentation using deep learning techniques. However, when applied to new conditions, these applications frequently face the difficulty of overfitting, resulting in lower segmentation performance. In the context of potato farming, where diseases have a large influence on yields, it is critical for the agricultural economy to quickly and properly identify these diseases. Traditional data augmentation approaches, such as rotation, flip, and translation, have limitations and frequently fail to provide strong generalization results. To address these issues, our research employs a novel approach termed as PotatoGANs. In this novel data augmentation approach, two types of Generative Adversarial Networks (GANs) are utilized to generate synthetic potato disease images from healthy potato images. This approach not only expands the dataset but also adds variety, which helps to enhance model generalization. Using the Inception score as a measure, our experiments show the better quality and realisticness of the images created by PotatoGANs, emphasizing their capacity to resemble real disease images closely. The CycleGAN model outperforms the Pix2Pix GAN model in terms of image quality, as evidenced by its higher IS scores CycleGAN achieves higher Inception scores (IS) of 1.2001 and 1.0900 for black scurf and common scab, respectively. This synthetic data can significantly improve the training of large neural networks. It also reduces data collection costs while enhancing data diversity and generalization capabilities. Our work improves interpretability by combining three gradient-based Explainable AI algorithms (GradCAM, GradCAM++, and ScoreCAM) with three distinct CNN architectures (DenseNet169, Resnet152 V2, InceptionResNet V2) for potato disease classification.", "date": "2024-05-12", "categories": []}
{"id": "2506.16318", "title": "Segment Anything for Satellite Imagery: A Strong Baseline and a Regional Dataset for Automatic Field Delineation", "abstract": "Accurate mapping of agricultural field boundaries is essential for the efficient operation of agriculture. Automatic extraction from high-resolution satellite imagery, supported by computer vision techniques, can avoid costly ground surveys. In this paper, we present a pipeline for field delineation based on the Segment Anything Model (SAM), introducing a fine-tuning strategy to adapt SAM to this task. In addition to using published datasets, we describe a method for acquiring a complementary regional dataset that covers areas beyond current sources. Extensive experiments assess segmentation accuracy and evaluate the generalization capabilities. Our approach provides a robust baseline for automated field delineation. The new regional dataset, known as ERAS, is now publicly available.", "date": "2025-06-19", "categories": []}
{"id": "2505.16033", "title": "An Approach Towards Identifying Bangladeshi Leaf Diseases through Transfer Learning and XAI", "abstract": "Leaf diseases are harmful conditions that affect the health, appearance and productivity of plants, leading to significant plant loss and negatively impacting farmers' livelihoods. These diseases cause visible symptoms such as lesions, color changes, and texture variations, making it difficult for farmers to manage plant health, especially in large or remote farms where expert knowledge is limited. The main motivation of this study is to provide an efficient and accessible solution for identifying plant leaf diseases in Bangladesh, where agriculture plays a critical role in food security. The objective of our research is to classify 21 distinct leaf diseases across six plants using deep learning models, improving disease detection accuracy while reducing the need for expert involvement. Deep Learning (DL) techniques, including CNN and Transfer Learning (TL) models like VGG16, VGG19, MobileNetV2, InceptionV3, ResNet50V2 and Xception are used. VGG19 and Xception achieve the highest accuracies, with 98.90% and 98.66% respectively. Additionally, Explainable AI (XAI) techniques such as GradCAM, GradCAM++, LayerCAM, ScoreCAM and FasterScoreCAM are used to enhance transparency by highlighting the regions of the models focused on during disease classification. This transparency ensures that farmers can understand the model's predictions and take necessary action. This approach not only improves disease management but also supports farmers in making informed decisions, leading to better plant protection and increased agricultural productivity.", "date": "2025-05-21", "categories": []}
{"id": "2506.12295", "title": "MatchPlant: An Open-Source Pipeline for UAV-Based Single-Plant Detection and Data Extraction", "abstract": "Accurate identification of individual plants from unmanned aerial vehicle (UAV) images is essential for advancing high-throughput phenotyping and supporting data-driven decision-making in plant breeding. This study presents MatchPlant, a modular, graphical user interface-supported, open-source Python pipeline for UAV-based single-plant detection and geospatial trait extraction. MatchPlant enables end-to-end workflows by integrating UAV image processing, user-guided annotation, Convolutional Neural Network model training for object detection, forward projection of bounding boxes onto an orthomosaic, and shapefile generation for spatial phenotypic analysis. In an early-season maize case study, MatchPlant achieved reliable detection performance (validation AP: 89.6%, test AP: 85.9%) and effectively projected bounding boxes, covering 89.8% of manually annotated boxes with 87.5% of projections achieving an Intersection over Union (IoU) greater than 0.5. Trait values extracted from predicted bounding instances showed high agreement with manual annotations (r = 0.87-0.97, IoU >= 0.4). Detection outputs were reused across time points to extract plant height and Normalized Difference Vegetation Index with minimal additional annotation, facilitating efficient temporal phenotyping. By combining modular design, reproducibility, and geospatial precision, MatchPlant offers a scalable framework for UAV-based plant-level analysis with broad applicability in agricultural and environmental monitoring.", "date": "2025-06-14", "categories": []}
{"id": "2506.13335", "title": "Advancing Image-Based Grapevine Variety Classification with a New Benchmark and Evaluation of Masked Autoencoders", "abstract": "Grapevine varieties are essential for the economies of many wine-producing countries, influencing the production of wine, juice, and the consumption of fruits and leaves. Traditional identification methods, such as ampelography and molecular analysis, have limitations: ampelography depends on expert knowledge and is inherently subjective, while molecular methods are costly and time-intensive. To address these limitations, recent studies have applied deep learning (DL) models to classify grapevine varieties using image data. However, due to the small dataset sizes, these methods often depend on transfer learning from datasets from other domains, e.g., ImageNet1K (IN1K), which can lead to performance degradation due to domain shift and supervision collapse. In this context, self-supervised learning (SSL) methods can be a good tool to avoid this performance degradation, since they can learn directly from data, without external labels. This study presents an evaluation of Masked Autoencoders (MAEs) for identifying grapevine varieties based on field-acquired images. The main contributions of this study include two benchmarks comprising 43 grapevine varieties collected across different seasons, an analysis of MAE's application in the agricultural context, and a performance comparison of trained models across seasons. Our results show that a ViT-B/16 model pre-trained with MAE and the unlabeled dataset achieved an F1 score of 0.7956, outperforming all other models. Additionally, we observed that pre-trained models benefit from long pre-training, perform well under low-data training regime, and that simple data augmentation methods are more effective than complex ones. The study also found that the mask ratio in MAE impacts performance only marginally.", "date": "2025-06-16", "categories": []}
{"id": "2412.08477", "title": "Accurate Water Level Monitoring in AWD Rice Cultivation Using Convolutional Neural Networks", "abstract": "The Alternate Wetting and Drying (AWD) method is a rice-growing water management technique promoted as a sustainable alternative to Continuous Flooding (CF). Climate change has placed the agricultural sector in a challenging position, particularly as global water resources become increasingly scarce, affecting rice production on irrigated lowlands. Rice, a staple food for over half of the world's population, demands significantly more water than other major crops. In Bangladesh, Boro rice, in particular, requires considerable water inputs during its cultivation. Traditionally, farmers manually measure water levels, a process that is both time-consuming and prone to errors. While ultrasonic sensors offer improvements in water height measurement, they still face limitations, such as susceptibility to weather conditions and environmental factors. To address these issues, we propose a novel approach that automates water height measurement using computer vision, specifically through a convolutional neural network (CNN). Our attention-based architecture achieved an $R^2$ score of 0.9885 and a Mean Squared Error (MSE) of 0.2766, providing a more accurate and efficient solution for managing AWD systems.", "date": "2024-12-11", "categories": []}
{"id": "2506.06155", "title": "Fine-grained Hierarchical Crop Type Classification from Integrated Hyperspectral EnMAP Data and Multispectral Sentinel-2 Time Series: A Large-scale Dataset and Dual-stream Transformer Method", "abstract": "Fine-grained crop type classification serves as the fundamental basis for large-scale crop mapping and plays a vital role in ensuring food security. It requires simultaneous capture of both phenological dynamics (obtained from multi-temporal satellite data like Sentinel-2) and subtle spectral variations (demanding nanometer-scale spectral resolution from hyperspectral imagery). Research combining these two modalities remains scarce currently due to challenges in hyperspectral data acquisition and crop types annotation costs. To address these issues, we construct a hierarchical hyperspectral crop dataset (H2Crop) by integrating 30m-resolution EnMAP hyperspectral data with Sentinel-2 time series. With over one million annotated field parcels organized in a four-tier crop taxonomy, H2Crop establishes a vital benchmark for fine-grained agricultural crop classification and hyperspectral image processing. We propose a dual-stream Transformer architecture that synergistically processes these modalities. It coordinates two specialized pathways: a spectral-spatial Transformer extracts fine-grained signatures from hyperspectral EnMAP data, while a temporal Swin Transformer extracts crop growth patterns from Sentinel-2 time series. The designed hierarchical classification head with hierarchical fusion then simultaneously delivers multi-level crop type classification across all taxonomic tiers. Experiments demonstrate that adding hyperspectral EnMAP data to Sentinel-2 time series yields a 4.2% average F1-scores improvement (peaking at 6.3%). Extensive comparisons also confirm our method's higher accuracy over existing deep learning approaches for crop type classification and the consistent benefits of hyperspectral data across varying temporal windows and crop change scenarios. Codes and dataset are available at https://github.com/flyakon/H2Crop.", "date": "2025-06-06", "categories": []}
{"id": "2506.07357", "title": "CBAM-STN-TPS-YOLO: Enhancing Agricultural Object Detection through Spatially Adaptive Attention Mechanisms", "abstract": "Object detection is vital in precision agriculture for plant monitoring, disease detection, and yield estimation. However, models like YOLO struggle with occlusions, irregular structures, and background noise, reducing detection accuracy. While Spatial Transformer Networks (STNs) improve spatial invariance through learned transformations, affine mappings are insufficient for non-rigid deformations such as bent leaves and overlaps.\n  We propose CBAM-STN-TPS-YOLO, a model integrating Thin-Plate Splines (TPS) into STNs for flexible, non-rigid spatial transformations that better align features. Performance is further enhanced by the Convolutional Block Attention Module (CBAM), which suppresses background noise and emphasizes relevant spatial and channel-wise features.\n  On the occlusion-heavy Plant Growth and Phenotyping (PGP) dataset, our model outperforms STN-YOLO in precision, recall, and mAP. It achieves a 12% reduction in false positives, highlighting the benefits of improved spatial flexibility and attention-guided refinement. We also examine the impact of the TPS regularization parameter in balancing transformation smoothness and detection performance.\n  This lightweight model improves spatial awareness and supports real-time edge deployment, making it ideal for smart farming applications requiring accurate and efficient monitoring.", "date": "2025-06-09", "categories": []}
{"id": "2502.01850", "title": "Foundation Model-Based Apple Ripeness and Size Estimation for Selective Harvesting", "abstract": "Harvesting is a critical task in the tree fruit industry, demanding extensive manual labor and substantial costs, and exposing workers to potential hazards. Recent advances in automated harvesting offer a promising solution by enabling efficient, cost-effective, and ergonomic fruit picking within tight harvesting windows. However, existing harvesting technologies often indiscriminately harvest all visible and accessible fruits, including those that are unripe or undersized. This study introduces a novel foundation model-based framework for efficient apple ripeness and size estimation. Specifically, we curated two public RGBD-based Fuji apple image datasets, integrating expanded annotations for ripeness (\"Ripe\" vs. \"Unripe\") based on fruit color and image capture dates. The resulting comprehensive dataset, Fuji-Ripeness-Size Dataset, includes 4,027 images and 16,257 annotated apples with ripeness and size labels. Using Grounding-DINO, a language-model-based object detector, we achieved robust apple detection and ripeness classification, outperforming other state-of-the-art models. Additionally, we developed and evaluated six size estimation algorithms, selecting the one with the lowest error and variation for optimal performance. The Fuji-Ripeness-Size Dataset and the apple detection and size estimation algorithms are made publicly available, which provides valuable benchmarks for future studies in automated and selective harvesting.", "date": "2025-02-03", "categories": []}
{"id": "2503.06608", "title": "GroMo: Plant Growth Modeling with Multiview Images", "abstract": "Understanding plant growth dynamics is essential for applications in agriculture and plant phenotyping. We present the Growth Modelling (GroMo) challenge, which is designed for two primary tasks: (1) plant age prediction and (2) leaf count estimation, both essential for crop monitoring and precision agriculture. For this challenge, we introduce GroMo25, a dataset with images of four crops: radish, okra, wheat, and mustard. Each crop consists of multiple plants (p1, p2, ..., pn) captured over different days (d1, d2, ..., dm) and categorized into five levels (L1, L2, L3, L4, L5). Each plant is captured from 24 different angles with a 15-degree gap between images. Participants are required to perform both tasks for all four crops with these multiview images. We proposed a Multiview Vision Transformer (MVVT) model for the GroMo challenge and evaluated the crop-wise performance on GroMo25. MVVT reports an average MAE of 7.74 for age prediction and an MAE of 5.52 for leaf count. The GroMo Challenge aims to advance plant phenotyping research by encouraging innovative solutions for tracking and predicting plant growth. The GitHub repository is publicly available at https://github.com/mriglab/GroMo-Plant-Growth-Modeling-with-Multiview-Images.", "date": "2025-03-09", "categories": []}
{"id": "2506.06084", "title": "WisWheat: A Three-Tiered Vision-Language Dataset for Wheat Management", "abstract": "Wheat management strategies play a critical role in determining yield. Traditional management decisions often rely on labour-intensive expert inspections, which are expensive, subjective and difficult to scale. Recently, Vision-Language Models (VLMs) have emerged as a promising solution to enable scalable, data-driven management support. However, due to a lack of domain-specific knowledge, directly applying VLMs to wheat management tasks results in poor quantification and reasoning capabilities, ultimately producing vague or even misleading management recommendations. In response, we propose WisWheat, a wheat-specific dataset with a three-layered design to enhance VLM performance on wheat management tasks: (1) a foundational pretraining dataset of 47,871 image-caption pairs for coarsely adapting VLMs to wheat morphology; (2) a quantitative dataset comprising 7,263 VQA-style image-question-answer triplets for quantitative trait measuring tasks; and (3) an Instruction Fine-tuning dataset with 4,888 samples targeting biotic and abiotic stress diagnosis and management plan for different phenological stages. Extensive experimental results demonstrate that fine-tuning open-source VLMs (e.g., Qwen2.5 7B) on our dataset leads to significant performance improvements. Specifically, the Qwen2.5 VL 7B fine-tuned on our wheat instruction dataset achieves accuracy scores of 79.2% and 84.6% on wheat stress and growth stage conversation tasks respectively, surpassing even general-purpose commercial models such as GPT-4o by a margin of 11.9% and 34.6%.", "date": "2025-06-06", "categories": []}
{"id": "2505.08302", "title": "Knowledge-Informed Deep Learning for Irrigation Type Mapping from Remote Sensing", "abstract": "Accurate mapping of irrigation methods is crucial for sustainable agricultural practices and food systems. However, existing models that rely solely on spectral features from satellite imagery are ineffective due to the complexity of agricultural landscapes and limited training data, making this a challenging problem. We present Knowledge-Informed Irrigation Mapping (KIIM), a novel Swin-Transformer based approach that uses (i) a specialized projection matrix to encode crop to irrigation probability, (ii) a spatial attention map to identify agricultural lands from non-agricultural lands, (iii) bi-directional cross-attention to focus complementary information from different modalities, and (iv) a weighted ensemble for combining predictions from images and crop information. Our experimentation on five states in the US shows up to 22.9\\% (IoU) improvement over baseline with a 71.4% (IoU) improvement for hard-to-classify drip irrigation. In addition, we propose a two-phase transfer learning approach to enhance cross-state irrigation mapping, achieving a 51% IoU boost in a state with limited labeled data. The ability to achieve baseline performance with only 40% of the training data highlights its efficiency, reducing the dependency on extensive manual labeling efforts and making large-scale, automated irrigation mapping more feasible and cost-effective.", "date": "2025-05-13", "categories": []}
{"id": "2506.00735", "title": "Involution-Infused DenseNet with Two-Step Compression for Resource-Efficient Plant Disease Classification", "abstract": "Agriculture is vital for global food security, but crops are vulnerable to diseases that impact yield and quality. While Convolutional Neural Networks (CNNs) accurately classify plant diseases using leaf images, their high computational demands hinder their deployment in resource-constrained settings such as smartphones, edge devices, and real-time monitoring systems. This study proposes a two-step model compression approach integrating Weight Pruning and Knowledge Distillation, along with the hybridization of DenseNet with Involutional Layers. Pruning reduces model size and computational load, while distillation improves the smaller student models performance by transferring knowledge from a larger teacher network. The hybridization enhances the models ability to capture spatial features efficiently. These compressed models are suitable for real-time applications, promoting precision agriculture through rapid disease identification and crop management. The results demonstrate ResNet50s superior performance post-compression, achieving 99.55% and 98.99% accuracy on the PlantVillage and PaddyLeaf datasets, respectively. The DenseNet-based model, optimized for efficiency, recorded 99.21% and 93.96% accuracy with a minimal parameter count. Furthermore, the hybrid model achieved 98.87% and 97.10% accuracy, supporting the practical deployment of energy-efficient devices for timely disease intervention and sustainable farming practices.", "date": "2025-05-31", "categories": []}
{"id": "2405.07550", "title": "Wild Berry image dataset collected in Finnish forests and peatlands using drones", "abstract": "Berry picking has long-standing traditions in Finland, yet it is challenging and can potentially be dangerous. The integration of drones equipped with advanced imaging techniques represents a transformative leap forward, optimising harvests and promising sustainable practices. We propose WildBe, the first image dataset of wild berries captured in peatlands and under the canopy of Finnish forests using drones. Unlike previous and related datasets, WildBe includes new varieties of berries, such as bilberries, cloudberries, lingonberries, and crowberries, captured under severe light variations and in cluttered environments. WildBe features 3,516 images, including a total of 18,468 annotated bounding boxes. We carry out a comprehensive analysis of WildBe using six popular object detectors, assessing their effectiveness in berry detection across different forest regions and camera types. WildBe is publicly available on HuggingFace at https://huggingface.co/datasets/FBK-TeV/WildBe.", "date": "2024-05-13", "categories": []}
{"id": "2505.21544", "title": "Vision Meets Language: A RAG-Augmented YOLOv8 Framework for Coffee Disease Diagnosis and Farmer Assistance", "abstract": "As a social being, we have an intimate bond with the environment. A plethora of things in human life, such as lifestyle, health, and food are dependent on the environment and agriculture. It comes under our responsibility to support the environment as well as agriculture. However, traditional farming practices often result in inefficient resource use and environmental challenges. To address these issues, precision agriculture has emerged as a promising approach that leverages advanced technologies to optimise agricultural processes. In this work, a hybrid approach is proposed that combines the three different potential fields of model AI: object detection, large language model (LLM), and Retrieval-Augmented Generation (RAG). In this novel framework, we have tried to combine the vision and language models to work together to identify potential diseases in the tree leaf. This study introduces a novel AI-based precision agriculture system that uses Retrieval Augmented Generation (RAG) to provide context-aware diagnoses and natural language processing (NLP) and YOLOv8 for crop disease detection. The system aims to tackle major issues with large language models (LLMs), especially hallucinations and allows for adaptive treatment plans and real-time disease detection. The system provides an easy-to-use interface to the farmers, which they can use to detect the different diseases related to coffee leaves by just submitting the image of the affected leaf the model will detect the diseases as well as suggest potential remediation methodologies which aim to lower the use of pesticides, preserving livelihoods, and encouraging environmentally friendly methods. With an emphasis on scalability, dependability, and user-friendliness, the project intends to improve RAG-integrated object detection systems for wider agricultural applications in the future.", "date": "2025-05-24", "categories": []}
{"id": "2505.21558", "title": "A Novel Convolutional Neural Network-Based Framework for Complex Multiclass Brassica Seed Classification", "abstract": "Agricultural research has accelerated in recent years, yet farmers often lack the time and resources for on-farm research due to the demands of crop production and farm operations. Seed classification offers valuable insights into quality control, production efficiency, and impurity detection. Early identification of seed types is critical to reducing the cost and risk associated with field emergence, which can lead to yield losses or disruptions in downstream processes like harvesting. Seed sampling supports growers in monitoring and managing seed quality, improving precision in determining seed purity levels, guiding management adjustments, and enhancing yield estimations. This study proposes a novel convolutional neural network (CNN)-based framework for the efficient classification of ten common Brassica seed types. The approach addresses the inherent challenge of texture similarity in seed images using a custom-designed CNN architecture. The model's performance was evaluated against several pre-trained state-of-the-art architectures, with adjustments to layer configurations for optimized classification. Experimental results using our collected Brassica seed dataset demonstrate that the proposed model achieved a high accuracy rate of 93 percent.", "date": "2025-05-26", "categories": []}
{"id": "2505.21746", "title": "Learning to See More: UAS-Guided Super-Resolution of Satellite Imagery for Precision Agriculture", "abstract": "Unmanned Aircraft Systems (UAS) and satellites are key data sources for precision agriculture, yet each presents trade-offs. Satellite data offer broad spatial, temporal, and spectral coverage but lack the resolution needed for many precision farming applications, while UAS provide high spatial detail but are limited by coverage and cost, especially for hyperspectral data. This study presents a novel framework that fuses satellite and UAS imagery using super-resolution methods. By integrating data across spatial, spectral, and temporal domains, we leverage the strengths of both platforms cost-effectively. We use estimation of cover crop biomass and nitrogen (N) as a case study to evaluate our approach. By spectrally extending UAS RGB data to the vegetation red edge and near-infrared regions, we generate high-resolution Sentinel-2 imagery and improve biomass and N estimation accuracy by 18% and 31%, respectively. Our results show that UAS data need only be collected from a subset of fields and time points. Farmers can then 1) enhance the spectral detail of UAS RGB imagery; 2) increase the spatial resolution by using satellite data; and 3) extend these enhancements spatially and across the growing season at the frequency of the satellite flights. Our SRCNN-based spectral extension model shows considerable promise for model transferability over other cropping systems in the Upper and Lower Chesapeake Bay regions. Additionally, it remains effective even when cloud-free satellite data are unavailable, relying solely on the UAS RGB input. The spatial extension model produces better biomass and N predictions than models built on raw UAS RGB images. Once trained with targeted UAS RGB data, the spatial extension model allows farmers to stop repeated UAS flights. While we introduce super-resolution advances, the core contribution is a lightweight and scalable system for affordable on-farm use.", "date": "2025-05-27", "categories": []}
{"id": "2505.19863", "title": "FruitNeRF++: A Generalized Multi-Fruit Counting Method Utilizing Contrastive Learning and Neural Radiance Fields", "abstract": "We introduce FruitNeRF++, a novel fruit-counting approach that combines contrastive learning with neural radiance fields to count fruits from unstructured input photographs of orchards. Our work is based on FruitNeRF, which employs a neural semantic field combined with a fruit-specific clustering approach. The requirement for adaptation for each fruit type limits the applicability of the method, and makes it difficult to use in practice. To lift this limitation, we design a shape-agnostic multi-fruit counting framework, that complements the RGB and semantic data with instance masks predicted by a vision foundation model. The masks are used to encode the identity of each fruit as instance embeddings into a neural instance field. By volumetrically sampling the neural fields, we extract a point cloud embedded with the instance features, which can be clustered in a fruit-agnostic manner to obtain the fruit count. We evaluate our approach using a synthetic dataset containing apples, plums, lemons, pears, peaches, and mangoes, as well as a real-world benchmark apple dataset. Our results demonstrate that FruitNeRF++ is easier to control and compares favorably to other state-of-the-art methods.", "date": "2025-05-26", "categories": []}
{"id": "2505.07444", "title": "Lightweight Multispectral Crop-Weed Segmentation for Precision Agriculture", "abstract": "Efficient crop-weed segmentation is critical for site-specific weed control in precision agriculture. Conventional CNN-based methods struggle to generalize and rely on RGB imagery, limiting performance under complex field conditions. To address these challenges, we propose a lightweight transformer-CNN hybrid. It processes RGB, Near-Infrared (NIR), and Red-Edge (RE) bands using specialized encoders and dynamic modality integration. Evaluated on the WeedsGalore dataset, the model achieves a segmentation accuracy (mean IoU) of 78.88%, outperforming RGB-only models by 15.8 percentage points. With only 8.7 million parameters, the model offers high accuracy, computational efficiency, and potential for real-time deployment on Unmanned Aerial Vehicles (UAVs) and edge devices, advancing precision weed management.", "date": "2025-05-12", "categories": []}
{"id": "2505.14029", "title": "AppleGrowthVision: A large-scale stereo dataset for phenological analysis, fruit detection, and 3D reconstruction in apple orchards", "abstract": "Deep learning has transformed computer vision for precision agriculture, yet apple orchard monitoring remains limited by dataset constraints. The lack of diverse, realistic datasets and the difficulty of annotating dense, heterogeneous scenes. Existing datasets overlook different growth stages and stereo imagery, both essential for realistic 3D modeling of orchards and tasks like fruit localization, yield estimation, and structural analysis. To address these gaps, we present AppleGrowthVision, a large-scale dataset comprising two subsets. The first includes 9,317 high resolution stereo images collected from a farm in Brandenburg (Germany), covering six agriculturally validated growth stages over a full growth cycle. The second subset consists of 1,125 densely annotated images from the same farm in Brandenburg and one in Pillnitz (Germany), containing a total of 31,084 apple labels. AppleGrowthVision provides stereo-image data with agriculturally validated growth stages, enabling precise phenological analysis and 3D reconstructions. Extending MinneApple with our data improves YOLOv8 performance by 7.69 % in terms of F1-score, while adding it to MinneApple and MAD boosts Faster R-CNN F1-score by 31.06 %. Additionally, six BBCH stages were predicted with over 95 % accuracy using VGG16, ResNet152, DenseNet201, and MobileNetv2. AppleGrowthVision bridges the gap between agricultural science and computer vision, by enabling the development of robust models for fruit detection, growth modeling, and 3D analysis in precision agriculture. Future work includes improving annotation, enhancing 3D reconstruction, and extending multimodal analysis across all growth stages.", "date": "2025-05-20", "categories": []}
{"id": "2308.05257", "title": "Developing a Hybrid Convolutional Neural Network for Automatic Aphid Counting in Sugar Beet Fields", "abstract": "Aphids can cause direct damage and indirect virus transmission to crops. Timely monitoring and control of their populations are thus critical. However, the manual counting of aphids, which is the most common practice, is labor-intensive and time-consuming. Additionally, two of the biggest challenges in aphid counting are that aphids are small objects and their density distributions are varied in different areas of the field. To address these challenges, we proposed a hybrid automatic aphid counting network architecture which integrates the detection network and the density map estimation network. When the distribution density of aphids is low, it utilizes an improved Yolov5 to count aphids. Conversely, when the distribution density of aphids is high, it switches to CSRNet to count aphids. To the best of our knowledge, this is the first framework integrating the detection network and the density map estimation network for counting tasks. Through comparison experiments of counting aphids, it verified that our proposed approach outperforms all other methods in counting aphids. It achieved the lowest MAE and RMSE values for both the standard and high-density aphid datasets: 2.93 and 4.01 (standard), and 34.19 and 38.66 (high-density), respectively. Moreover, the AP of the improved Yolov5 is 5% higher than that of the original Yolov5. Especially for extremely small aphids and densely distributed aphids, the detection performance of the improved Yolov5 is significantly better than the original Yolov5. This work provides an effective early warning caused by aphids in sugar beet fields, offering protection for sugar beet growth and ensuring sugar beet yield. The datasets and project code are released at: https://github.com/JunfengGaolab/Counting-Aphids.", "date": "2023-08-09", "categories": []}
{"id": "2505.10770", "title": "Geofenced Unmanned Aerial Robotic Defender for Deer Detection and Deterrence (GUARD)", "abstract": "Wildlife-induced crop damage, particularly from deer, threatens agricultural productivity. Traditional deterrence methods often fall short in scalability, responsiveness, and adaptability to diverse farmland environments. This paper presents an integrated unmanned aerial vehicle (UAV) system designed for autonomous wildlife deterrence, developed as part of the Farm Robotics Challenge. Our system combines a YOLO-based real-time computer vision module for deer detection, an energy-efficient coverage path planning algorithm for efficient field monitoring, and an autonomous charging station for continuous operation of the UAV. In collaboration with a local Minnesota farmer, the system is tailored to address practical constraints such as terrain, infrastructure limitations, and animal behavior. The solution is evaluated through a combination of simulation and field testing, demonstrating robust detection accuracy, efficient coverage, and extended operational time. The results highlight the feasibility and effectiveness of drone-based wildlife deterrence in precision agriculture, offering a scalable framework for future deployment and extension.", "date": "2025-05-16", "categories": []}
{"id": "2505.11020", "title": "Classifying Shelf Life Quality of Pineapples by Combining Audio and Visual Features", "abstract": "Determining the shelf life quality of pineapples using non-destructive methods is a crucial step to reduce waste and increase income. In this paper, a multimodal and multiview classification model was constructed to classify pineapples into four quality levels based on audio and visual characteristics. For research purposes, we compiled and released the PQC500 dataset consisting of 500 pineapples with two modalities: one was tapping pineapples to record sounds by multiple microphones and the other was taking pictures by multiple cameras at different locations, providing multimodal and multi-view audiovisual features. We modified the contrastive audiovisual masked autoencoder to train the cross-modal-based classification model by abundant combinations of audio and visual pairs. In addition, we proposed to sample a compact size of training data for efficient computation. The experiments were evaluated under various data and model configurations, and the results demonstrated that the proposed cross-modal model trained using audio-major sampling can yield 84% accuracy, outperforming the unimodal models of only audio and only visual by 6% and 18%, respectively.", "date": "2025-05-16", "categories": []}
{"id": "2505.08537", "title": "The RaspGrade Dataset: Towards Automatic Raspberry Ripeness Grading with Deep Learning", "abstract": "This research investigates the application of computer vision for rapid, accurate, and non-invasive food quality assessment, focusing on the novel challenge of real-time raspberry grading into five distinct classes within an industrial environment as the fruits move along a conveyor belt. To address this, a dedicated dataset of raspberries, namely RaspGrade, was acquired and meticulously annotated. Instance segmentation experiments revealed that accurate fruit-level masks can be obtained; however, the classification of certain raspberry grades presents challenges due to color similarities and occlusion, while others are more readily distinguishable based on color. The acquired and annotated RaspGrade dataset is accessible on Hugging Face at: https://huggingface.co/datasets/FBK-TeV/RaspGrade.", "date": "2025-05-13", "categories": []}
{"id": "2505.01823", "title": "PhytoSynth: Leveraging Multi-modal Generative Models for Crop Disease Data Generation with Novel Benchmarking and Prompt Engineering Approach", "abstract": "Collecting large-scale crop disease images in the field is labor-intensive and time-consuming. Generative models (GMs) offer an alternative by creating synthetic samples that resemble real-world images. However, existing research primarily relies on Generative Adversarial Networks (GANs)-based image-to-image translation and lack a comprehensive analysis of computational requirements in agriculture. Therefore, this research explores a multi-modal text-to-image approach for generating synthetic crop disease images and is the first to provide computational benchmarking in this context. We trained three Stable Diffusion (SD) variants-SDXL, SD3.5M (medium), and SD3.5L (large)-and fine-tuned them using Dreambooth and Low-Rank Adaptation (LoRA) fine-tuning techniques to enhance generalization. SD3.5M outperformed the others, with an average memory usage of 18 GB, power consumption of 180 W, and total energy use of 1.02 kWh/500 images (0.002 kWh per image) during inference task. Our results demonstrate SD3.5M's ability to generate 500 synthetic images from just 36 in-field samples in 1.5 hours. We recommend SD3.5M for efficient crop disease data generation.", "date": "2025-05-03", "categories": []}
{"id": "2505.07840", "title": "Evaluation of UAV-Based RGB and Multispectral Vegetation Indices for Precision Agriculture in Palm Tree Cultivation", "abstract": "Precision farming relies on accurate vegetation monitoring to enhance crop productivity and promote sustainable agricultural practices. This study presents a comprehensive evaluation of UAV-based imaging for vegetation health assessment in a palm tree cultivation region in Dubai. By comparing multispectral and RGB image data, we demonstrate that RGBbased vegetation indices offer performance comparable to more expensive multispectral indices, providing a cost-effective alternative for large-scale agricultural monitoring. Using UAVs equipped with multispectral sensors, indices such as NDVI and SAVI were computed to categorize vegetation into healthy, moderate, and stressed conditions. Simultaneously, RGB-based indices like VARI and MGRVI delivered similar results in vegetation classification and stress detection. Our findings highlight the practical benefits of integrating RGB imagery into precision farming, reducing operational costs while maintaining accuracy in plant health monitoring. This research underscores the potential of UAVbased RGB imaging as a powerful tool for precision agriculture, enabling broader adoption of data-driven decision-making in crop management. By leveraging the strengths of both multispectral and RGB imaging, this work advances the state of UAV applications in agriculture, paving the way for more efficient and scalable farming solutions.", "date": "2025-05-06", "categories": []}
{"id": "2505.00741", "title": "Detection and Classification of Diseases in Multi-Crop Leaves using LSTM\n  and CNN Models", "abstract": "  Plant diseases pose a serious challenge to agriculture by reducing crop yield\nand affecting food quality. Early detection and classification of these\ndiseases are essential for minimising losses and improving crop management\npractices. This study applies Convolutional Neural Networks (CNN) and Long\nShort-Term Memory (LSTM) models to classify plant leaf diseases using a dataset\ncontaining 70,295 training images and 17,572 validation images across 38\ndisease classes. The CNN model was trained using the Adam optimiser with a\nlearning rate of 0.0001 and categorical cross-entropy as the loss function.\nAfter 10 training epochs, the model achieved a training accuracy of 99.1% and a\nvalidation accuracy of 96.4%. The LSTM model reached a validation accuracy of\n93.43%. Performance was evaluated using precision, recall, F1-score, and\nconfusion matrix, confirming the reliability of the CNN-based approach. The\nresults suggest that deep learning models, particularly CNN, enable an\neffective solution for accurate and scalable plant disease classification,\nsupporting practical applications in agricultural monitoring.\n", "date": "2025-04-30", "categories": []}
{"id": "2505.00995", "title": "Optimizing Indoor Farm Monitoring Efficiency Using UAV: Yield Estimation\n  in a GNSS-Denied Cherry Tomato Greenhouse", "abstract": "  As the agricultural workforce declines and labor costs rise, robotic yield\nestimation has become increasingly important. While unmanned ground vehicles\n(UGVs) are commonly used for indoor farm monitoring, their deployment in\ngreenhouses is often constrained by infrastructure limitations, sensor\nplacement challenges, and operational inefficiencies. To address these issues,\nwe develop a lightweight unmanned aerial vehicle (UAV) equipped with an RGB-D\ncamera, a 3D LiDAR, and an IMU sensor. The UAV employs a LiDAR-inertial\nodometry algorithm for precise navigation in GNSS-denied environments and\nutilizes a 3D multi-object tracking algorithm to estimate the count and weight\nof cherry tomatoes. We evaluate the system using two dataset: one from a\nharvesting row and another from a growing row. In the harvesting-row dataset,\nthe proposed system achieves 94.4\\% counting accuracy and 87.5\\% weight\nestimation accuracy within a 13.2-meter flight completed in 10.5 seconds. For\nthe growing-row dataset, which consists of occluded unripened fruits, we\nqualitatively analyze tracking performance and highlight future research\ndirections for improving perception in greenhouse with strong occlusions. Our\nfindings demonstrate the potential of UAVs for efficient robotic yield\nestimation in commercial greenhouses.\n", "date": "2025-05-02", "categories": []}
{"id": "2411.15923", "title": "Deep Learning for automated multi-scale functional field boundaries\n  extraction using multi-date Sentinel-2 and PlanetScope imagery: Case Study of\n  Netherlands and Pakistan", "abstract": "  This study explores the effectiveness of multi-temporal satellite imagery for\nbetter functional field boundary delineation using deep learning semantic\nsegmentation architecture on two distinct geographical and multi-scale farming\nsystems of Netherlands and Pakistan. Multidate images of April, August and\nOctober 2022 were acquired for PlanetScope and Sentinel-2 in sub regions of\nNetherlands and November 2022, February and March 2023 for selected area of\nDunyapur in Pakistan. For Netherlands, Basic registration crop parcels (BRP)\nvector layer was used as labeled training data. while self-crafted field\nboundary vector data were utilized for Pakistan. Four deep learning models with\nUNET architecture were evaluated using different combinations of multi-date\nimages and NDVI stacks in the Netherlands subregions. A comparative analysis of\nIoU scores assessed the effectiveness of the proposed multi-date NDVI stack\napproach. These findings were then applied for transfer learning, using\npre-trained models from the Netherlands on the selected area in Pakistan.\nAdditionally, separate models were trained using self-crafted field boundary\ndata for Pakistan, and combined models were developed using data from both the\nNetherlands and Pakistan. Results indicate that multi-date NDVI stacks provide\nadditional temporal context, reflecting crop growth over different times of the\nseason. The study underscores the critical role of multi-scale ground\ninformation from diverse geographical areas in developing robust and\nuniversally applicable models for field boundary delineation. The results also\nhighlight the importance of fine spatial resolution for extraction of field\nboundaries in regions with small scale framing. The findings can be extended to\nmulti-scale implementations for improved automatic field boundary delineation\nin heterogeneous agricultural environments.\n", "date": "2024-11-24", "categories": []}
{"id": "2501.02921", "title": "Unsupervised Tomato Split Anomaly Detection using Hyperspectral Imaging\n  and Variational Autoencoders", "abstract": "  Tomato anomalies/damages pose a significant challenge in greenhouse farming.\nWhile this method of cultivation benefits from efficient resource utilization,\nanomalies can significantly degrade the quality of farm produce. A common\nanomaly associated with tomatoes is splitting, characterized by the development\nof cracks on the tomato skin, which degrades its quality. Detecting this type\nof anomaly is challenging due to dynamic variations in appearance and sizes,\ncompounded by dataset scarcity. We address this problem in an unsupervised\nmanner by utilizing a tailored variational autoencoder (VAE) with hyperspectral\ninput. Preliminary analysis of the dataset enabled us to select the optimal\nrange of wavelengths for detecting this anomaly. Our findings indicate that the\n530nm - 550nm range is suitable for identifying tomato dry splits. The proposed\nVAE model achieved a 97% detection accuracy for tomato split anomalies in the\ntest data. The analysis on reconstruction loss allow us to not only detect the\nanomalies but also to some degree estimate the anomalous regions.\n", "date": "2025-01-06", "categories": []}
{"id": "2504.19334", "title": "Enhancing seeding efficiency using a computer vision system to monitor\n  furrow quality in real-time", "abstract": "  Effective seed sowing in precision agriculture is hindered by challenges such\nas residue accumulation, low soil temperatures, and hair pinning (crop residue\npushed in the trench by furrow opener), which obstruct optimal trench\nformation. Row cleaners are employed to mitigate these issues, but there is a\nlack of quantitative methods to assess trench cleanliness. In this study, a\nnovel computer vision-based method was developed to evaluate row cleaner\nperformance. Multiple air seeders were equipped with a video acquisition system\nto capture trench conditions after row cleaner operation, enabling an effective\ncomparison of the performance of each row cleaner. The captured data were used\nto develop a segmentation model that analyzed key elements such as soil, straw,\nand machinery. Using the results from the segmentation model, an objective\nmethod was developed to quantify row cleaner performance. The results\ndemonstrated the potential of this method to improve row cleaner selection and\nenhance seeding efficiency in precision agriculture.\n", "date": "2025-04-27", "categories": []}
{"id": "2504.17943", "title": "Predicting Dairy Calf Body Weight from Depth Images Using Deep Learning\n  (YOLOv8) and Threshold Segmentation with Cross-Validation and Longitudinal\n  Analysis", "abstract": "  Monitoring calf body weight (BW) before weaning is essential for assessing\ngrowth, feed efficiency, health, and weaning readiness. However, labor, time,\nand facility constraints limit BW collection. Additionally, Holstein calf coat\npatterns complicate image-based BW estimation, and few studies have explored\nnon-contact measurements taken at early time points for predicting later BW.\nThe objectives of this study were to (1) develop deep learning-based\nsegmentation models for extracting calf body metrics, (2) compare deep learning\nsegmentation with threshold-based methods, and (3) evaluate BW prediction using\nsingle-time-point cross-validation with linear regression (LR) and extreme\ngradient boosting (XGBoost) and multiple-time-point cross-validation with LR,\nXGBoost, and a linear mixed model (LMM). Depth images from Holstein (n = 63)\nand Jersey (n = 5) pre-weaning calves were collected, with 20 Holstein calves\nbeing weighed manually. Results showed that You Only Look Once version 8\n(YOLOv8) deep learning segmentation (intersection over union = 0.98)\noutperformed threshold-based methods (0.89). In single-time-point\ncross-validation, XGBoost achieved the best BW prediction (R^2 = 0.91, mean\nabsolute percentage error (MAPE) = 4.37%), while LMM provided the most accurate\nlongitudinal BW prediction (R^2 = 0.99, MAPE = 2.39%). These findings highlight\nthe potential of deep learning for automated BW prediction, enhancing farm\nmanagement.\n", "date": "2025-04-24", "categories": []}
{"id": "2504.16658", "title": "A Time Series Dataset of NIR Spectra and RGB and NIR-HSI Images of the\n  Barley Germination Process", "abstract": "  We provide an open-source dataset of RGB and NIR-HSI (near-infrared\nhyperspectral imaging) images with associated segmentation masks and NIR\nspectra of 2242 individual malting barley kernels. We imaged every kernel\npre-exposure to moisture and every 24 hours after exposure to moisture for five\nconsecutive days. Every barley kernel was labeled as germinated or not\ngerminated during each image acquisition. The barley kernels were imaged with\nblack filter paper as the background, facilitating straight-forward intensity\nthreshold-based segmentation, e.g., by Otsu's method. This dataset facilitates\ntime series analysis of germination time for barley kernels using either RGB\nimage analysis, NIR spectral analysis, NIR-HSI analysis, or a combination\nhereof.\n", "date": "2025-04-23", "categories": []}
{"id": "2504.12992", "title": "Enhancing Cocoa Pod Disease Classification via Transfer Learning and\n  Ensemble Methods: Toward Robust Predictive Modeling", "abstract": "  This study presents an ensemble-based approach for cocoa pod disease\nclassification by integrating transfer learning with three ensemble learning\nstrategies: Bagging, Boosting, and Stacking. Pre-trained convolutional neural\nnetworks, including VGG16, VGG19, ResNet50, ResNet101, InceptionV3, and\nXception, were fine-tuned and employed as base learners to detect three disease\ncategories: Black Pod Rot, Pod Borer, and Healthy. A balanced dataset of 6,000\ncocoa pod images was curated and augmented to ensure robustness against\nvariations in lighting, orientation, and disease severity. The performance of\neach ensemble method was evaluated using accuracy, precision, recall, and\nF1-score. Experimental results show that Bagging consistently achieved superior\nclassification performance with a test accuracy of 100%, outperforming Boosting\n(97%) and Stacking (92%). The findings confirm that combining transfer learning\nwith ensemble techniques improves model generalization and reliability, making\nit a promising direction for precision agriculture and automated crop disease\nmanagement.\n", "date": "2025-04-17", "categories": []}
{"id": "2504.13099", "title": "RF-DETR Object Detection vs YOLOv12 : A Study of Transformer-based and\n  CNN-based Architectures for Single-Class and Multi-Class Greenfruit Detection\n  in Complex Orchard Environments Under Label Ambiguity", "abstract": "  This study conducts a detailed comparison of RF-DETR object detection base\nmodel and YOLOv12 object detection model configurations for detecting\ngreenfruits in a complex orchard environment marked by label ambiguity,\nocclusions, and background blending. A custom dataset was developed featuring\nboth single-class (greenfruit) and multi-class (occluded and non-occluded\ngreenfruits) annotations to assess model performance under dynamic real-world\nconditions. RF-DETR object detection model, utilizing a DINOv2 backbone and\ndeformable attention, excelled in global context modeling, effectively\nidentifying partially occluded or ambiguous greenfruits. In contrast, YOLOv12\nleveraged CNN-based attention for enhanced local feature extraction, optimizing\nit for computational efficiency and edge deployment. RF-DETR achieved the\nhighest mean Average Precision (mAP50) of 0.9464 in single-class detection,\nproving its superior ability to localize greenfruits in cluttered scenes.\nAlthough YOLOv12N recorded the highest mAP@50:95 of 0.7620, RF-DETR\nconsistently outperformed in complex spatial scenarios. For multi-class\ndetection, RF-DETR led with an mAP@50 of 0.8298, showing its capability to\ndifferentiate between occluded and non-occluded fruits, while YOLOv12L scored\nhighest in mAP@50:95 with 0.6622, indicating better classification in detailed\nocclusion contexts. Training dynamics analysis highlighted RF-DETR's swift\nconvergence, particularly in single-class settings where it plateaued within 10\nepochs, demonstrating the efficiency of transformer-based architectures in\nadapting to dynamic visual data. These findings validate RF-DETR's\neffectiveness for precision agricultural applications, with YOLOv12 suited for\nfast-response scenarios. >Index Terms: RF-DETR object detection, YOLOv12,\nYOLOv13, YOLOv14, YOLOv15, YOLOE, YOLO World, YOLO, You Only Look Once,\nRoboflow, Detection Transformers, CNNs\n", "date": "2025-04-17", "categories": []}
{"id": "2411.16219", "title": "Weakly Supervised Panoptic Segmentation for Defect-Based Grading of\n  Fresh Produce", "abstract": "  Visual inspection for defect grading in agricultural supply chains is crucial\nbut traditionally labor-intensive and error-prone. Automated computer vision\nmethods typically require extensively annotated datasets, which are often\nunavailable in decentralized supply chains. We address this challenge by\nevaluating the Segment Anything Model (SAM) to generate dense panoptic\nsegmentation masks from sparse annotations. These dense predictions are then\nused to train a supervised panoptic segmentation model. Focusing on banana\nsurface defects (bruises and scars), we validate our approach using 476 field\nimages annotated with 1440 defects. While SAM-generated masks generally align\nwith human annotations, substantially reducing annotation effort, we explicitly\nidentify failure cases associated with specific defect sizes and shapes.\nDespite these limitations, our approach offers practical estimates of defect\nnumber and relative size from panoptic masks, underscoring the potential and\ncurrent boundaries of foundation models for defect quantification in low-data\nagricultural scenarios. GitHub:\nhttps://github.com/manuelknott/banana-defect-segmentation\n", "date": "2024-11-25", "categories": []}
{"id": "2504.08568", "title": "Banana Ripeness Level Classification using a Simple CNN Model Trained\n  with Real and Synthetic Datasets", "abstract": "  The level of ripeness is essential in determining the quality of bananas. To\ncorrectly estimate banana maturity, the metrics of international marketing\nstandards need to be considered. However, the process of assessing the maturity\nof bananas at an industrial level is still carried out using manual methods.\nThe use of CNN models is an attractive tool to solve the problem, but there is\na limitation regarding the availability of sufficient data to train these\nmodels reliably. On the other hand, in the state-of-the-art, existing CNN\nmodels and the available data have reported that the accuracy results are\nacceptable in identifying banana maturity. For this reason, this work presents\nthe generation of a robust dataset that combines real and synthetic data for\ndifferent levels of banana ripeness. In addition, it proposes a simple CNN\narchitecture, which is trained with synthetic data and using the transfer\nlearning technique, the model is improved to classify real data, managing to\ndetermine the level of maturity of the banana. The proposed CNN model is\nevaluated with several architectures, then hyper-parameter configurations are\nvaried, and optimizers are used. The results show that the proposed CNN model\nreaches a high accuracy of 0.917 and a fast execution time.\n", "date": "2025-04-11", "categories": []}
{"id": "2501.15262", "title": "TflosYOLO+TFSC: An Accurate and Robust Model for Estimating Flower Count\n  and Flowering Period", "abstract": "  Tea flowers play a crucial role in taxonomic research and hybrid breeding for\nthe tea plant. As traditional methods of observing tea flower traits are\nlabor-intensive and inaccurate, we propose TflosYOLO and TFSC model for tea\nflowering quantifying, which enable estimation of flower count and flowering\nperiod. In this study, a highly representative and diverse dataset was\nconstructed by collecting flower images from 29 tea accessions in 2 years.\nBased on this dataset, the TflosYOLO model was built on the YOLOv5 architecture\nand enhanced with the Squeeze-and-Excitation (SE) network, which is the first\nmodel to offer a viable solution for detecting and counting tea flowers. The\nTflosYOLO model achieved an mAP50 of 0.874, outperforming YOLOv5, YOLOv7 and\nYOLOv8. Furthermore, TflosYOLO model was tested on 34 datasets encompassing 26\ntea accessions, five flowering stages, various lighting conditions, and pruned\n/ unpruned plants, demonstrating high generalization and robustness. The\ncorrelation coefficient (R^2) between the predicted and actual flower counts\nwas 0.974. Additionally, the TFSC (Tea Flowering Stage Classification) model, a\n7-layer neural network was designed for automatic classification of the\nflowering period. TFSC model was evaluated on 2 years and achieved an accuracy\nof 0.738 and 0.899 respectively. Using the TflosYOLO+TFSC model, we monitored\nthe tea flowering dynamics and tracked the changes in flowering stages across\nvarious tea accessions. The framework provides crucial support for tea plant\nbreeding programs and phenotypic analysis of germplasm resources.\n", "date": "2025-01-25", "categories": []}
{"id": "2504.02880", "title": "Global Rice Multi-Class Segmentation Dataset (RiceSEG): A Comprehensive\n  and Diverse High-Resolution RGB-Annotated Images for the Development and\n  Benchmarking of Rice Segmentation Algorithms", "abstract": "  Developing computer vision-based rice phenotyping techniques is crucial for\nprecision field management and accelerating breeding, thereby continuously\nadvancing rice production. Among phenotyping tasks, distinguishing image\ncomponents is a key prerequisite for characterizing plant growth and\ndevelopment at the organ scale, enabling deeper insights into eco-physiological\nprocesses. However, due to the fine structure of rice organs and complex\nillumination within the canopy, this task remains highly challenging,\nunderscoring the need for a high-quality training dataset. Such datasets are\nscarce, both due to a lack of large, representative collections of rice field\nimages and the time-intensive nature of annotation. To address this gap, we\nestablished the first comprehensive multi-class rice semantic segmentation\ndataset, RiceSEG. We gathered nearly 50,000 high-resolution, ground-based\nimages from five major rice-growing countries (China, Japan, India, the\nPhilippines, and Tanzania), encompassing over 6,000 genotypes across all growth\nstages. From these original images, 3,078 representative samples were selected\nand annotated with six classes (background, green vegetation, senescent\nvegetation, panicle, weeds, and duckweed) to form the RiceSEG dataset. Notably,\nthe sub-dataset from China spans all major genotypes and rice-growing\nenvironments from the northeast to the south. Both state-of-the-art\nconvolutional neural networks and transformer-based semantic segmentation\nmodels were used as baselines. While these models perform reasonably well in\nsegmenting background and green vegetation, they face difficulties during the\nreproductive stage, when canopy structures are more complex and multiple\nclasses are involved. These findings highlight the importance of our dataset\nfor developing specialized segmentation models for rice and other crops.\n", "date": "2025-04-02", "categories": []}
{"id": "2504.03510", "title": "FADConv: A Frequency-Aware Dynamic Convolution for Farmland\n  Non-agriculturalization Identification and Segmentation", "abstract": "  Cropland non-agriculturalization refers to the conversion of arable land into\nnon-agricultural uses such as forests, residential areas, and construction\nsites. This phenomenon not only directly leads to the loss of cropland\nresources but also poses systemic threats to food security and agricultural\nsustainability. Accurate identification of cropland and non-cropland areas is\ncrucial for detecting and addressing this issue. Traditional CNNs employ static\nconvolution layers, while dynamic convolution studies demonstrate that\nadaptively weighting multiple convolutional kernels through attention\nmechanisms can enhance accuracy. However, existing dynamic convolution methods\nrelying on Global Average Pooling (GAP) for attention weight allocation suffer\nfrom information loss, limiting segmentation precision. This paper proposes\nFrequency-Aware Dynamic Convolution (FADConv) and a Frequency Attention (FAT)\nmodule to address these limitations. Building upon the foundational structure\nof dynamic convolution, we designed FADConv by integrating 2D Discrete Cosine\nTransform (2D DCT) to capture frequency domain features and fuse them. FAT\nmodule generates high-quality attention weights that replace the traditional\nGAP method,making the combination between dynamic convolution kernels more\nreasonable.Experiments on the GID and Hi-CNA datasets demonstrate that FADConv\nsignificantly improves segmentation accuracy with minimal computational\noverhead. For instance, ResNet18 with FADConv achieves 1.9% and 2.7% increases\nin F1-score and IoU for cropland segmentation on GID, with only 58.87M\nadditional MAdds. Compared to other dynamic convolution approaches, FADConv\nexhibits superior performance in cropland segmentation tasks.\n", "date": "2025-04-04", "categories": []}
{"id": "2504.02534", "title": "Delineate Anything: Resolution-Agnostic Field Boundary Delineation on\n  Satellite Imagery", "abstract": "  The accurate delineation of agricultural field boundaries from satellite\nimagery is vital for land management and crop monitoring. However, current\nmethods face challenges due to limited dataset sizes, resolution discrepancies,\nand diverse environmental conditions. We address this by reformulating the task\nas instance segmentation and introducing the Field Boundary Instance\nSegmentation - 22M dataset (FBIS-22M), a large-scale, multi-resolution dataset\ncomprising 672,909 high-resolution satellite image patches (ranging from 0.25 m\nto 10 m) and 22,926,427 instance masks of individual fields, significantly\nnarrowing the gap between agricultural datasets and those in other computer\nvision domains. We further propose Delineate Anything, an instance segmentation\nmodel trained on our new FBIS-22M dataset. Our proposed model sets a new\nstate-of-the-art, achieving a substantial improvement of 88.5% in mAP@0.5 and\n103% in mAP@0.5:0.95 over existing methods, while also demonstrating\nsignificantly faster inference and strong zero-shot generalization across\ndiverse image resolutions and unseen geographic regions. Code, pre-trained\nmodels, and the FBIS-22M dataset are available at\nhttps://lavreniuk.github.io/Delineate-Anything.\n", "date": "2025-04-03", "categories": []}
{"id": "2404.14653", "title": "Machine Vision-Based Assessment of Fall Color Changes and its\n  Relationship with Leaf Nitrogen Concentration", "abstract": "  Apple(\\textit{Malus domestica} Borkh.) trees are deciduous, shedding leaves\neach year. This process is preceded by a gradual change in leaf color from\ngreen to yellow as chlorophyll is degraded prior to abscission. The initiation\nand rate of this color change are affected by many factors including leaf\nnitrogen (N) concentration. We predict that leaf color during this transition\nmay be indicative of the nitrogen status of apple trees. This study assesses a\nmachine vision-based system for quantifying the change in leaf color and its\ncorrelation with leaf nitrogen content. An image dataset was collected in color\nand 3D over five weeks in the fall of 2021 and 2023 at a commercial orchard\nusing a ground vehicle-based stereovision sensor. Trees in the foreground were\nsegmented from the point cloud using color and depth thresholding methods.\nThen, to estimate the proportion of yellow leaves per canopy, the color\ninformation of the segmented canopy area was quantified using a custom-defined\nmetric, \\textit{yellowness index} (a normalized ratio of yellow to green\nfoliage in the tree) that varied from -1 to +1 (-1 being completely green and\n+1 being completely yellow). Both K-means-based methods and gradient boosting\nmethods were used to estimate the \\textit{yellowness index}. The gradient\nboosting based method proposed in this study was better than the K-means-based\nmethod (both in terms of computational time and accuracy), achieving an $R^2$\nof 0.72 in estimating the \\textit{yellowness index}. The metric was able to\ncapture the gradual color transition from green to yellow over the study\nduration. Trees with lower leaf nitrogen showed the color transition to yellow\nearlier than the trees with higher nitrogen.\n  Keywords: Fruit Tree Nitrogen Management, Machine Vision, Point Cloud\nSegmentation, Precision Nitrogen Management\n", "date": "2024-04-23", "categories": []}
{"id": "2410.19846", "title": "YOLO11 and Vision Transformers based 3D Pose Estimation of Immature\n  Green Fruits in Commercial Apple Orchards for Robotic Thinning", "abstract": "  In this study, a robust method for 3D pose estimation of immature green\napples (fruitlets) in commercial orchards was developed, utilizing the\nYOLO11(or YOLOv11) object detection and pose estimation algorithm alongside\nVision Transformers (ViT) for depth estimation (Dense Prediction Transformer\n(DPT) and Depth Anything V2). For object detection and pose estimation,\nperformance comparisons of YOLO11 (YOLO11n, YOLO11s, YOLO11m, YOLO11l and\nYOLO11x) and YOLOv8 (YOLOv8n, YOLOv8s, YOLOv8m, YOLOv8l and YOLOv8x) were made\nunder identical hyperparameter settings among the all configurations. It was\nobserved that YOLO11n surpassed all configurations of YOLO11 and YOLOv8 in\nterms of box precision and pose precision, achieving scores of 0.91 and 0.915,\nrespectively. Conversely, YOLOv8n exhibited the highest box and pose recall\nscores of 0.905 and 0.925, respectively. Regarding the mean average precision\nat 50\\% intersection over union (mAP@50), YOLO11s led all configurations with a\nbox mAP@50 score of 0.94, while YOLOv8n achieved the highest pose mAP@50 score\nof 0.96. In terms of image processing speed, YOLO11n outperformed all\nconfigurations with an impressive inference speed of 2.7 ms, significantly\nfaster than the quickest YOLOv8 configuration, YOLOv8n, which processed images\nin 7.8 ms. Subsequent integration of ViTs for the green fruit's pose depth\nestimation revealed that Depth Anything V2 outperformed Dense Prediction\nTransformer in 3D pose length validation, achieving the lowest Root Mean Square\nError (RMSE) of 1.52 and Mean Absolute Error (MAE) of 1.28, demonstrating\nexceptional precision in estimating immature green fruit lengths. Integration\nof YOLO11 and Depth Anything Model provides a promising solution to 3D pose\nestimation of immature green fruits for robotic thinning applications. (YOLOv11\npose detection, YOLOv11 Pose, YOLOv11 Keypoints detection, YOLOv11 pose\nestimation)\n", "date": "2024-10-21", "categories": []}
{"id": "2503.24032", "title": "BBoxCut: A Targeted Data Augmentation Technique for Enhancing Wheat Head\n  Detection Under Occlusions", "abstract": "  Wheat plays a critical role in global food security, making it one of the\nmost extensively studied crops. Accurate identification and measurement of key\ncharacteristics of wheat heads are essential for breeders to select varieties\nfor cross-breeding, with the goal of developing nutrient-dense, resilient, and\nsustainable cultivars. Traditionally, these measurements are performed\nmanually, which is both time-consuming and inefficient. Advances in digital\ntechnologies have paved the way for automating this process. However, field\nconditions pose significant challenges, such as occlusions of leaves,\noverlapping wheat heads, varying lighting conditions, and motion blur. In this\npaper, we propose a novel data augmentation technique, BBoxCut, which uses\nrandom localized masking to simulate occlusions caused by leaves and\nneighboring wheat heads. We evaluated our approach using three state-of-the-art\nobject detectors and observed mean average precision (mAP) gains of 2.76, 3.26,\nand 1.9 for Faster R-CNN, FCOS, and DETR, respectively. Our augmentation\ntechnique led to significant improvements both qualitatively and\nquantitatively. In particular, the improvements were particularly evident in\nscenarios involving occluded wheat heads, demonstrating the robustness of our\nmethod in challenging field conditions.\n", "date": "2025-03-31", "categories": []}
{"id": "2503.21990", "title": "AgRowStitch: A High-fidelity Image Stitching Pipeline for Ground-based\n  Agricultural Images", "abstract": "  Agricultural imaging often requires individual images to be stitched together\ninto a final mosaic for analysis. However, agricultural images can be\nparticularly challenging to stitch because feature matching across images is\ndifficult due to repeated textures, plants are non-planar, and mosaics built\nfrom many images can accumulate errors that cause drift. Although these issues\ncan be mitigated by using georeferenced images or taking images at high\naltitude, there is no general solution for images taken close to the crop. To\naddress this, we created a user-friendly and open source pipeline for stitching\nground-based images of a linear row of crops that does not rely on additional\ndata. First, we use SuperPoint and LightGlue to extract and match features\nwithin small batches of images. Then we stitch the images in each batch in\nseries while imposing constraints on the camera movement. After straightening\nand rescaling each batch mosaic, all batch mosaics are stitched together in\nseries and then straightened into a final mosaic. We tested the pipeline on\nimages collected along 72 m long rows of crops using two different agricultural\nrobots and a camera manually carried over the row. In all three cases, the\npipeline produced high-quality mosaics that could be used to georeference real\nworld positions with a mean absolute error of 20 cm. This approach provides\naccessible leaf-scale stitching to users who need to coarsely georeference\npositions within a row, but do not have access to accurate positional data or\nsophisticated imaging systems.\n", "date": "2025-03-27", "categories": []}
{"id": "2503.21250", "title": "Orange Quality Grading with Deep Learning", "abstract": "  Orange grading is a crucial step in the fruit industry, as it helps to sort\noranges according to different criteria such as size, quality, ripeness, and\nhealth condition, ensuring safety for human consumption and better price\nallocation and client satisfaction. Automated grading enables faster\nprocessing, precision, and reduced human labor. In this paper, we implement a\ndeep learning-based solution for orange grading via machine vision. Unlike\ntypical grading systems that analyze fruits from a single view, we capture\nmultiview images of each single orange in order to enable a richer\nrepresentation. Afterwards, we compose the acquired images into one collage.\nThis enables the analysis of the whole orange skin. We train a convolutional\nneural network (CNN) on the composed images to grade the oranges into three\nclasses, namely good, bad, and undefined. We also evaluate the performance with\ntwo different CNNs (ResNet-18 and SqueezeNet). We show experimentally that\nmulti-view grading is superior to single view grading.\n", "date": "2025-03-27", "categories": []}
{"id": "2503.21323", "title": "DuckSegmentation: A segmentation model based on the AnYue Hemp Duck\n  Dataset", "abstract": "  The modernization of smart farming is a way to improve agricultural\nproduction efficiency, and improve the agricultural production environment.\nAlthough many large models have achieved high accuracy in the task of object\nrecognition and segmentation, they cannot really be put into use in the farming\nindustry due to their own poor interpretability and limitations in\ncomputational volume. In this paper, we built AnYue Shelduck Dateset, which\ncontains a total of 1951 Shelduck datasets, and performed target detection and\nsegmentation annotation with the help of professional annotators. Based on\nAnYue ShelduckDateset, this paper describes DuckProcessing, an efficient and\npowerful module for duck identification based on real shelduckfarms. First of\nall, using the YOLOv8 module designed to divide the mahjong between them,\nPrecision reached 98.10%, Recall reached 96.53% and F1 score reached 0.95 on\nthe test set. Again using the DuckSegmentation segmentation model,\nDuckSegmentation reached 96.43% mIoU. Finally, the excellent DuckSegmentation\nwas used as the teacher model, and through knowledge distillation, Deeplabv3\nr50 was used as the student model, and the final student model achieved 94.49%\nmIoU on the test set. The method provides a new way of thinking in practical\nsisal duck smart farming.\n", "date": "2025-03-27", "categories": []}
{"id": "2503.20068", "title": "iNatAg: Multi-Class Classification Models Enabled by a Large-Scale\n  Benchmark Dataset with 4.7M Images of 2,959 Crop and Weed Species", "abstract": "  Accurate identification of crop and weed species is critical for precision\nagriculture and sustainable farming. However, it remains a challenging task due\nto a variety of factors -- a high degree of visual similarity among species,\nenvironmental variability, and a continued lack of large, agriculture-specific\nimage data. We introduce iNatAg, a large-scale image dataset which contains\nover 4.7 million images of 2,959 distinct crop and weed species, with precise\nannotations along the taxonomic hierarchy from binary crop/weed labels to\nspecific species labels. Curated from the broader iNaturalist database, iNatAg\ncontains data from every continent and accurately reflects the variability of\nnatural image captures and environments. Enabled by this data, we train\nbenchmark models built upon the Swin Transformer architecture and evaluate the\nimpact of various modifications such as the incorporation of geospatial data\nand LoRA finetuning. Our best models achieve state-of-the-art performance\nacross all taxonomic classification tasks, achieving 92.38\\% on crop and weed\nclassification. Furthermore, the scale of our dataset enables us to explore\nincorrect misclassifications and unlock new analytic possiblities for plant\nspecies. By combining large-scale species coverage, multi-task labels, and\ngeographic diversity, iNatAg provides a new foundation for building robust,\ngeolocation-aware agricultural classification systems. We release the iNatAg\ndataset publicly through AgML (https://github.com/Project-AgML/AgML), enabling\ndirect access and integration into agricultural machine learning workflows.\n", "date": "2025-03-25", "categories": []}
{"id": "2503.20419", "title": "Cherry Yield Forecast: Harvest Prediction for Individual Sweet Cherry\n  Trees", "abstract": "  This paper is part of a publication series from the For5G project that has\nthe goal of creating digital twins of sweet cherry trees. At the beginning a\nbrief overview of the revious work in this project is provided. Afterwards the\nfocus shifts to a crucial problem in the fruit farming domain: the difficulty\nof making reliable yield predictions early in the season. Following three Satin\nsweet cherry trees along the year 2023 enabled the collection of accurate\nground truth data about the development of cherries from dormancy until\nharvest. The methodology used to collect this data is presented, along with its\nvaluation and visualization. The predictive power of counting objects at all\nrelevant vegetative stages of the fruit development cycle in cherry trees with\nregards to yield predictions is investigated. It is found that all investigated\nfruit states are suitable for yield predictions based on linear regression.\nConceptionally, there is a trade-off between earliness and external events with\nthe potential to invalidate the prediction. Considering this, two optimal\ntimepoints are suggested that are opening cluster stage before the start of the\nflowering and the early fruit stage right after the second fruit drop. However,\nboth timepoints are challenging to solve with automated procedures based on\nimage data. Counting developing cherries based on images is exceptionally\ndifficult due to the small fruit size and their tendency to be occluded by\nleaves. It was not possible to obtain satisfying results relying on a\nstate-of-the-art fruit-counting method. Counting the elements within a bursting\nbud is also challenging, even when using high resolution cameras. It is\nconcluded that accurate yield prediction for sweet cherry trees is possible\nwhen objects are manually counted and that automated features extraction with\nsimilar accuracy remains an open problem yet to be solved.\n", "date": "2025-03-26", "categories": []}
{"id": "2307.01530", "title": "Tomato Maturity Recognition with Convolutional Transformers", "abstract": "  Tomatoes are a major crop worldwide, and accurately classifying their\nmaturity is important for many agricultural applications, such as harvesting,\ngrading, and quality control. In this paper, the authors propose a novel method\nfor tomato maturity classification using a convolutional transformer. The\nconvolutional transformer is a hybrid architecture that combines the strengths\nof convolutional neural networks (CNNs) and transformers. Additionally, this\nstudy introduces a new tomato dataset named KUTomaData, explicitly designed to\ntrain deep-learning models for tomato segmentation and classification.\nKUTomaData is a compilation of images sourced from a greenhouse in the UAE,\nwith approximately 700 images available for training and testing. The dataset\nis prepared under various lighting conditions and viewing perspectives and\nemploys different mobile camera sensors, distinguishing it from existing\ndatasets. The contributions of this paper are threefold:Firstly, the authors\npropose a novel method for tomato maturity classification using a modular\nconvolutional transformer. Secondly, the authors introduce a new tomato image\ndataset that contains images of tomatoes at different maturity levels. Lastly,\nthe authors show that the convolutional transformer outperforms\nstate-of-the-art methods for tomato maturity classification. The effectiveness\nof the proposed framework in handling cluttered and occluded tomato instances\nwas evaluated using two additional public datasets, Laboro Tomato and Rob2Pheno\nAnnotated Tomato, as benchmarks. The evaluation results across these three\ndatasets demonstrate the exceptional performance of our proposed framework,\nsurpassing the state-of-the-art by 58.14%, 65.42%, and 66.39% in terms of mean\naverage precision scores for KUTomaData, Laboro Tomato, and Rob2Pheno Annotated\nTomato, respectively.\n", "date": "2023-07-04", "categories": []}
{"id": "2503.00057", "title": "Improved YOLOv12 with LLM-Generated Synthetic Data for Enhanced Apple\n  Detection and Benchmarking Against YOLOv11 and YOLOv10", "abstract": "  This study evaluated the performance of the YOLOv12 object detection model,\nand compared against the performances YOLOv11 and YOLOv10 for apple detection\nin commercial orchards based on the model training completed entirely on\nsynthetic images generated by Large Language Models (LLMs). The YOLOv12n\nconfiguration achieved the highest precision at 0.916, the highest recall at\n0.969, and the highest mean Average Precision (mAP@50) at 0.978. In comparison,\nthe YOLOv11 series was led by YOLO11x, which achieved the highest precision at\n0.857, recall at 0.85, and mAP@50 at 0.91. For the YOLOv10 series, YOLOv10b and\nYOLOv10l both achieved the highest precision at 0.85, with YOLOv10n achieving\nthe highest recall at 0.8 and mAP@50 at 0.89. These findings demonstrated that\nYOLOv12, when trained on realistic LLM-generated datasets surpassed its\npredecessors in key performance metrics. The technique also offered a\ncost-effective solution by reducing the need for extensive manual data\ncollection in the agricultural field. In addition, this study compared the\ncomputational efficiency of all versions of YOLOv12, v11 and v10, where\nYOLOv11n reported the lowest inference time at 4.7 ms, compared to YOLOv12n's\n5.6 ms and YOLOv10n's 5.9 ms. Although YOLOv12 is new and more accurate than\nYOLOv11, and YOLOv10, YOLO11n still stays the fastest YOLO model among YOLOv10,\nYOLOv11 and YOLOv12 series of models. (Index: YOLOv12, YOLOv11, YOLOv10,\nYOLOv13, YOLOv14, YOLOv15, YOLOE, YOLO Object detection)\n", "date": "2025-02-26", "categories": []}
{"id": "2502.19351", "title": "Deep Learning-Based Transfer Learning for Classification of Cassava\n  Disease", "abstract": "  This paper presents a performance comparison among four Convolutional Neural\nNetwork architectures (EfficientNet-B3, InceptionV3, ResNet50, and VGG16) for\nclassifying cassava disease images. The images were sourced from an imbalanced\ndataset from a competition. Appropriate metrics were employed to address class\nimbalance. The results indicate that EfficientNet-B3 achieved on this task\naccuracy of 87.7%, precision of 87.8%, revocation of 87.8% and F1-Score of\n87.7%. These findings suggest that EfficientNet-B3 could be a valuable tool to\nsupport Digital Agriculture.\n", "date": "2025-02-26", "categories": []}
{"id": "2503.13778", "title": "Using 3D reconstruction from image motion to predict total leaf area in\n  dwarf tomato plants", "abstract": "  Accurate estimation of total leaf area (TLA) is crucial for evaluating plant\ngrowth, photosynthetic activity, and transpiration. However, it remains\nchallenging for bushy plants like dwarf tomatoes due to their complex canopies.\nTraditional methods are often labor-intensive, damaging to plants, or limited\nin capturing canopy complexity. This study evaluated a non-destructive method\ncombining sequential 3D reconstructions from RGB images and machine learning to\nestimate TLA for three dwarf tomato cultivars: Mohamed, Hahms Gelbe Topftomate,\nand Red Robin -- grown under controlled greenhouse conditions. Two experiments\n(spring-summer and autumn-winter) included 73 plants, yielding 418 TLA\nmeasurements via an \"onion\" approach. High-resolution videos were recorded, and\n500 frames per plant were used for 3D reconstruction. Point clouds were\nprocessed using four algorithms (Alpha Shape, Marching Cubes, Poisson's, Ball\nPivoting), and meshes were evaluated with seven regression models:\nMultivariable Linear Regression, Lasso Regression, Ridge Regression, Elastic\nNet Regression, Random Forest, Extreme Gradient Boosting, and Multilayer\nPerceptron. The Alpha Shape reconstruction ($\\alpha = 3$) with Extreme Gradient\nBoosting achieved the best performance ($R^2 = 0.80$, $MAE = 489 cm^2$).\nCross-experiment validation showed robust results ($R^2 = 0.56$, $MAE = 579\ncm^2$). Feature importance analysis identified height, width, and surface area\nas key predictors. This scalable, automated TLA estimation method is suited for\nurban farming and precision agriculture, offering applications in automated\npruning, resource efficiency, and sustainable food production. The approach\ndemonstrated robustness across variable environmental conditions and canopy\nstructures.\n", "date": "2025-03-17", "categories": []}
{"id": "2503.13080", "title": "Vision-based automatic fruit counting with UAV", "abstract": "  The use of unmanned aerial vehicles (UAVs) for smart agriculture is becoming\nincreasingly popular. This is evidenced by recent scientific works, as well as\nthe various competitions organised on this topic. Therefore, in this work we\npresent a system for automatic fruit counting using UAVs. To detect them, our\nsolution uses a vision algorithm that processes streams from an RGB camera and\na depth sensor using classical image operations. Our system also allows the\nplanning and execution of flight trajectories, taking into account the\nminimisation of flight time and distance covered. We tested the proposed\nsolution in simulation and obtained an average score of 87.27/100 points from a\ntotal of 500 missions. We also submitted it to the UAV Competition organised as\npart of the ICUAS 2024 conference, where we achieved an average score of\n84.83/100 points, placing 6th in a field of 23 teams and advancing to the\nfinals.\n", "date": "2025-03-17", "categories": []}
{"id": "2311.05029", "title": "S$^3$AD: Semi-supervised Small Apple Detection in Orchard Environments", "abstract": "  Crop detection is integral for precision agriculture applications such as\nautomated yield estimation or fruit picking. However, crop detection, e.g.,\napple detection in orchard environments remains challenging due to a lack of\nlarge-scale datasets and the small relative size of the crops in the image. In\nthis work, we address these challenges by reformulating the apple detection\ntask in a semi-supervised manner. To this end, we provide the large,\nhigh-resolution dataset MAD comprising 105 labeled images with 14,667 annotated\napple instances and 4,440 unlabeled images. Utilizing this dataset, we also\npropose a novel Semi-Supervised Small Apple Detection system S$^3$AD based on\ncontextual attention and selective tiling to improve the challenging detection\nof small apples, while limiting the computational overhead. We conduct an\nextensive evaluation on MAD and the MSU dataset, showing that S$^3$AD\nsubstantially outperforms strong fully-supervised baselines, including several\nsmall object detection systems, by up to $14.9\\%$. Additionally, we exploit the\ndetailed annotations of our dataset w.r.t. apple properties to analyze the\ninfluence of relative size or level of occlusion on the results of various\nsystems, quantifying current challenges.\n", "date": "2023-11-08", "categories": []}
{"id": "2503.10940", "title": "Automated Tomato Maturity Estimation Using an Optimized Residual Model\n  with Pruning and Quantization Techniques", "abstract": "  Tomato maturity plays a pivotal role in optimizing harvest timing and\nensuring product quality, but current methods struggle to achieve high accuracy\nalong computational efficiency simultaneously. Existing deep learning\napproaches, while accurate, are often too computationally demanding for\npractical use in resource-constrained agricultural settings. In contrast,\nsimpler techniques fail to capture the nuanced features needed for precise\nclassification. This study aims to develop a computationally efficient tomato\nclassification model using the ResNet-18 architecture optimized through\ntransfer learning, pruning, and quantization techniques. Our objective is to\naddress the dual challenge of maintaining high accuracy while enabling\nreal-time performance on low-power edge devices. Then, these models were\ndeployed on an edge device to investigate their performance for tomato maturity\nclassification. The quantized model achieved an accuracy of 97.81%, with an\naverage classification time of 0.000975 seconds per image. The pruned and\nauto-tuned model also demonstrated significant improvements in deployment\nmetrics, further highlighting the benefits of optimization techniques. These\nresults underscore the potential for a balanced solution that meets the\naccuracy and efficiency demands of modern agricultural production, paving the\nway for practical, real-world deployment in resource-limited environments.\n", "date": "2025-03-13", "categories": []}
{"id": "2503.08843", "title": "Keypoint Semantic Integration for Improved Feature Matching in Outdoor\n  Agricultural Environments", "abstract": "  Robust robot navigation in outdoor environments requires accurate perception\nsystems capable of handling visual challenges such as repetitive structures and\nchanging appearances. Visual feature matching is crucial to vision-based\npipelines but remains particularly challenging in natural outdoor settings due\nto perceptual aliasing. We address this issue in vineyards, where repetitive\nvine trunks and other natural elements generate ambiguous descriptors that\nhinder reliable feature matching. We hypothesise that semantic information tied\nto keypoint positions can alleviate perceptual aliasing by enhancing keypoint\ndescriptor distinctiveness. To this end, we introduce a keypoint semantic\nintegration technique that improves the descriptors in semantically meaningful\nregions within the image, enabling more accurate differentiation even among\nvisually similar local features. We validate this approach in two vineyard\nperception tasks: (i) relative pose estimation and (ii) visual localisation.\nAcross all tested keypoint types and descriptors, our method improves matching\naccuracy by 12.6%, demonstrating its effectiveness over multiple months in\nchallenging vineyard conditions.\n", "date": "2025-03-11", "categories": []}
{"id": "2211.07748", "title": "3D Reconstruction-Based Seed Counting of Sorghum Panicles for\n  Agricultural Inspection", "abstract": "  In this paper, we present a method for creating high-quality 3D models of\nsorghum panicles for phenotyping in breeding experiments. This is achieved with\na novel reconstruction approach that uses seeds as semantic landmarks in both\n2D and 3D. To evaluate the performance, we develop a new metric for assessing\nthe quality of reconstructed point clouds without having a ground-truth point\ncloud. Finally, a counting method is presented where the density of seed\ncenters in the 3D model allows 2D counts from multiple views to be effectively\ncombined into a whole-panicle count. We demonstrate that using this method to\nestimate seed count and weight for sorghum outperforms count extrapolation from\n2D images, an approach used in most state of the art methods for seeds and\ngrains of comparable size.\n", "date": "2022-11-14", "categories": []}
{"id": "2503.05568", "title": "TomatoScanner: phenotyping tomato fruit based on only RGB image", "abstract": "  In tomato greenhouse, phenotypic measurement is meaningful for researchers\nand farmers to monitor crop growth, thereby precisely control environmental\nconditions in time, leading to better quality and higher yield. Traditional\nphenotyping mainly relies on manual measurement, which is accurate but\ninefficient, more importantly, endangering the health and safety of people.\nSeveral studies have explored computer vision-based methods to replace manual\nphenotyping. However, the 2D-based need extra calibration, or cause destruction\nto fruit, or can only measure limited and meaningless traits. The 3D-based need\nextra depth camera, which is expensive and unacceptable for most farmers. In\nthis paper, we propose a non-contact tomato fruit phenotyping method, titled\nTomatoScanner, where RGB image is all you need for input. First, pixel feature\nis extracted by instance segmentation of our proposed EdgeYOLO with\npreprocessing of individual separation and pose correction. Second, depth\nfeature is extracted by depth estimation of Depth Pro. Third, pixel and depth\nfeature are fused to output phenotype results in reality. We establish\nself-built Tomato Phenotype Dataset to test TomatoScanner, which achieves\nexcellent phenotyping on width, height, vertical area and volume, with median\nrelative error of 5.63%, 7.03%, -0.64% and 37.06%, respectively. We propose and\nadd three innovative modules - EdgeAttention, EdgeLoss and EdgeBoost - into\nEdgeYOLO, to enhance the segmentation accuracy on edge portion. Precision and\nmean Edge Error greatly improve from 0.943 and 5.641% to 0.986 and 2.963%,\nrespectively. Meanwhile, EdgeYOLO keeps lightweight and efficient, with 48.7 M\nweights size and 76.34 FPS. Codes and datasets:\nhttps://github.com/AlexTraveling/TomatoScanner.\n", "date": "2025-03-07", "categories": []}
{"id": "2503.05630", "title": "Joint 3D Point Cloud Segmentation using Real-Sim Loop: From Panels to\n  Trees and Branches", "abstract": "  Modern orchards are planted in structured rows with distinct panel divisions\nto improve management. Accurate and efficient joint segmentation of point cloud\nfrom Panel to Tree and Branch (P2TB) is essential for robotic operations.\nHowever, most current segmentation methods focus on single instance\nsegmentation and depend on a sequence of deep networks to perform joint tasks.\nThis strategy hinders the use of hierarchical information embedded in the data,\nleading to both error accumulation and increased costs for annotation and\ncomputation, which limits its scalability for real-world applications. In this\nstudy, we proposed a novel approach that incorporated a Real2Sim L-TreeGen for\ntraining data generation and a joint model (J-P2TB) designed for the P2TB task.\nThe J-P2TB model, trained on the generated simulation dataset, was used for\njoint segmentation of real-world panel point clouds via zero-shot learning.\nCompared to representative methods, our model outperformed them in most\nsegmentation metrics while using 40% fewer learnable parameters. This Sim2Real\nresult highlighted the efficacy of L-TreeGen in model training and the\nperformance of J-P2TB for joint segmentation, demonstrating its strong\naccuracy, efficiency, and generalizability for real-world applications. These\nimprovements would not only greatly benefit the development of robots for\nautomated orchard operations but also advance digital twin technology.\n", "date": "2025-03-07", "categories": []}
{"id": "2503.03214", "title": "Rice Grain Size Measurement using Image Processing", "abstract": "  The rice grain quality can be determined from its size and chalkiness. The\ntraditional approach to measure the rice grain size involves manual inspection,\nwhich is inefficient and leads to inconsistent results. To address this issue,\nan image processing based approach is proposed and developed in this research.\nThe approach takes image of rice grains as input and outputs the number of rice\ngrains and size of each rice grain. The different steps, such as extraction of\nregion of interest, segmentation of rice grains, and sub-contours removal,\ninvolved in the proposed approach are discussed. The approach was tested on\nrice grain images captured from different height using mobile phone camera. The\nobtained results show that the proposed approach successfully detected 95\\% of\nthe rice grains and achieved 90\\% accuracy for length and width measurement.\n", "date": "2025-03-05", "categories": []}
{"id": "2502.20092", "title": "WalnutData: A UAV Remote Sensing Dataset of Green Walnuts and Model\n  Evaluation", "abstract": "  The UAV technology is gradually maturing and can provide extremely powerful\nsupport for smart agriculture and precise monitoring. Currently, there is no\ndataset related to green walnuts in the field of agricultural computer vision.\nThus, in order to promote the algorithm design in the field of agricultural\ncomputer vision, we used UAV to collect remote-sensing data from 8 walnut\nsample plots. Considering that green walnuts are subject to various lighting\nconditions and occlusion, we constructed a large-scale dataset with a\nhigher-granularity of target features - WalnutData. This dataset contains a\ntotal of 30,240 images and 706,208 instances, and there are 4 target\ncategories: being illuminated by frontal light and unoccluded (A1), being\nbacklit and unoccluded (A2), being illuminated by frontal light and occluded\n(B1), and being backlit and occluded (B2). Subsequently, we evaluated many\nmainstream algorithms on WalnutData and used these evaluation results as the\nbaseline standard. The dataset and all evaluation results can be obtained at\nhttps://github.com/1wuming/WalnutData.\n", "date": "2025-02-27", "categories": []}
{"id": "2503.02420", "title": "Exploring Model Quantization in GenAI-based Image Inpainting and\n  Detection of Arable Plants", "abstract": "  Deep learning-based weed control systems often suffer from limited training\ndata diversity and constrained on-board computation, impacting their real-world\nperformance. To overcome these challenges, we propose a framework that\nleverages Stable Diffusion-based inpainting to augment training data\nprogressively in 10% increments -- up to an additional 200%, thus enhancing\nboth the volume and diversity of samples. Our approach is evaluated on two\nstate-of-the-art object detection models, YOLO11(l) and RT-DETR(l), using the\nmAP50 metric to assess detection performance. We explore quantization\nstrategies (FP16 and INT8) for both the generative inpainting and detection\nmodels to strike a balance between inference speed and accuracy. Deployment of\nthe downstream models on the Jetson Orin Nano demonstrates the practical\nviability of our framework in resource-constrained environments, ultimately\nimproving detection accuracy and computational efficiency in intelligent weed\nmanagement systems.\n", "date": "2025-03-04", "categories": []}
{"id": "2502.20223", "title": "Deep Convolutional Neural Networks for Palm Fruit Maturity\n  Classification", "abstract": "  To maximize palm oil yield and quality, it is essential to harvest palm fruit\nat the optimal maturity stage. This project aims to develop an automated\ncomputer vision system capable of accurately classifying palm fruit images into\nfive ripeness levels. We employ deep Convolutional Neural Networks (CNNs) to\nclassify palm fruit images based on their maturity stage. A shallow CNN serves\nas the baseline model, while transfer learning and fine-tuning are applied to\npre-trained ResNet50 and InceptionV3 architectures. The study utilizes a\npublicly available dataset of over 8,000 images with significant variations,\nwhich is split into 80\\% for training and 20\\% for testing. The proposed deep\nCNN models achieve test accuracies exceeding 85\\% in classifying palm fruit\nmaturity stages. This research highlights the potential of deep learning for\nautomating palm fruit ripeness assessment, which can contribute to optimizing\nharvesting decisions and improving palm oil production efficiency.\n", "date": "2025-02-27", "categories": []}
{"id": "2501.10809", "title": "Efficient auto-labeling of large-scale poultry datasets (ALPD) using an\n  ensemble model with self- and active-learning approaches", "abstract": "  The rapid growth of artificial intelligence in poultry farming has\nhighlighted the challenge of efficiently labeling large, diverse datasets.\nManual annotation is time-consuming and costly, making it impractical for\nmodern systems that continuously generate data. This study addresses this\nchallenge by exploring semi-supervised auto-labeling methods, integrating self\nand active learning approaches to develop an efficient, label-scarce framework\nfor auto-labeling large poultry datasets (ALPD). For this study, video data\nwere collected from broilers and laying hens housed. Various machine learning\nmodels, including zero-shot models and supervised models, were utilized for\nbroilers and hens detection. The results showed that YOLOv8s-World and YOLOv9s\nperformed better when compared performance metrics for broiler and hen\ndetection under supervised learning, while among the semi-supervised model,\nYOLOv8s-ALPD achieved the highest precision (96.1%) and recall (99%) with an\nRMSE of 1.87. The hybrid YOLO-World model, incorporating the optimal YOLOv8s\nbackbone with zero-shot models, demonstrated the highest overall performance.\nIt achieved a precision of 99.2%, recall of 99.4%, and an F1 score of 98.7% for\ndetection. In addition, the semi-supervised models with minimal human\nintervention (active learning) reduced annotation time by over 80% compared to\nfull manual labeling. Moreover, integrating zero-shot models with the best\nmodels enhanced broiler and hen detection, achieving comparable results to\nsupervised models while significantly increasing speed. In conclusion,\nintegrating semi-supervised auto-labeling and zero-shot models significantly\nimproves detection accuracy. It reduces manual annotation efforts, offering a\npromising solution to optimize AI-driven systems in poultry farming, advancing\nprecision livestock management, and promoting more sustainable practices.\n", "date": "2025-01-18", "categories": []}
{"id": "2502.16950", "title": "A survey of datasets for computer vision in agriculture", "abstract": "  In agricultural research, there has been a recent surge in the amount of\nComputer Vision (CV) focused work. But unlike general CV research, large\nhigh-quality public datasets are sparsely available. This can be partially\nattributed to the high variability between different agricultural tasks, crops\nand environments as well as the complexity of data collection, but it is also\ninfluenced by the reticence to publish datasets by many authors. This, as well\nas the lack of a widely used agricultural data repository, are impactful\nfactors that hinder research in applied CV for agriculture as well as the usage\nof agricultural data in general-purpose CV research. In this survey, we provide\na large number of high-quality datasets of images taken on fields. Overall, we\nfind 45 datasets, which are listed in this paper as well as in an online\ncatalog on the project website:\nhttps://smartfarminglab.github.io/field_dataset_survey/.\n", "date": "2025-02-24", "categories": []}
{"id": "2502.14890", "title": "WeedVision: Multi-Stage Growth and Classification of Weeds using DETR\n  and RetinaNet for Precision Agriculture", "abstract": "  Weed management remains a critical challenge in agriculture, where weeds\ncompete with crops for essential resources, leading to significant yield\nlosses. Accurate detection of weeds at various growth stages is crucial for\neffective management yet challenging for farmers, as it requires identifying\ndifferent species at multiple growth phases. This research addresses these\nchallenges by utilizing advanced object detection models, specifically, the\nDetection Transformer (DETR) with a ResNet50 backbone and RetinaNet with a\nResNeXt101 backbone, to identify and classify 16 weed species of economic\nconcern across 174 classes, spanning their 11 weeks growth stages from seedling\nto maturity. A robust dataset comprising 203,567 images was developed,\nmeticulously labeled by species and growth stage. The models were rigorously\ntrained and evaluated, with RetinaNet demonstrating superior performance,\nachieving a mean Average Precision (mAP) of 0.907 on the training set and 0.904\non the test set, compared to DETR's mAP of 0.854 and 0.840, respectively.\nRetinaNet also outperformed DETR in recall and inference speed of 7.28 FPS,\nmaking it more suitable for real time applications. Both models showed improved\naccuracy as plants matured. This research provides crucial insights for\ndeveloping precise, sustainable, and automated weed management strategies,\npaving the way for real time species specific detection systems and advancing\nAI-assisted agriculture through continued innovation in model development and\nearly detection accuracy.\n", "date": "2025-02-16", "categories": []}
{"id": "2502.14360", "title": "Weed Detection using Convolutional Neural Network", "abstract": "  In this paper we use convolutional neural networks (CNNs) for weed detection\nin agricultural land. We specifically investigate the application of two CNN\nlayer types, Conv2d and dilated Conv2d, for weed detection in crop fields. The\nsuggested method extracts features from the input photos using pre-trained\nmodels, which are subsequently adjusted for weed detection. The findings of the\nexperiment, which used a sizable collection of dataset consisting of 15336\nsegments, being 3249 of soil, 7376 of soybean, 3520 grass and 1191 of broadleaf\nweeds. show that the suggested approach can accurately and successfully detect\nweeds at an accuracy of 94%. This study has significant ramifications for\nlowering the usage of toxic herbicides and increasing the effectiveness of weed\nmanagement in agriculture.\n", "date": "2025-02-20", "categories": []}
{"id": "2502.14455", "title": "An Efficient Ground-aerial Transportation System for Pest Control\n  Enabled by AI-based Autonomous Nano-UAVs", "abstract": "  Efficient crop production requires early detection of pest outbreaks and\ntimely treatments; we consider a solution based on a fleet of multiple\nautonomous miniaturized unmanned aerial vehicles (nano-UAVs) to visually detect\npests and a single slower heavy vehicle that visits the detected outbreaks to\ndeliver treatments. To cope with the extreme limitations aboard nano-UAVs,\ne.g., low-resolution sensors and sub-100 mW computational power budget, we\ndesign, fine-tune, and optimize a tiny image-based convolutional neural network\n(CNN) for pest detection. Despite the small size of our CNN (i.e., 0.58\nGOps/inference), on our dataset, it scores a mean average precision (mAP) of\n0.79 in detecting harmful bugs, i.e., 14% lower mAP but 32x fewer operations\nthan the best-performing CNN in the literature. Our CNN runs in real-time at\n6.8 frame/s, requiring 33 mW on a GWT GAP9 System-on-Chip aboard a Crazyflie\nnano-UAV. Then, to cope with in-field unexpected obstacles, we leverage a\nglobal+local path planner based on the A* algorithm. The global path planner\ndetermines the best route for the nano-UAV to sweep the entire area, while the\nlocal one runs up to 50 Hz aboard our nano-UAV and prevents collision by\nadjusting the short-distance path. Finally, we demonstrate with in-simulator\nexperiments that once a 25 nano-UAVs fleet has combed a 200x200 m vineyard,\ncollected information can be used to plan the best path for the tractor,\nvisiting all and only required hotspots. In this scenario, our efficient\ntransportation system, compared to a traditional single-ground vehicle\nperforming both inspection and treatment, can save up to 20 h working time.\n", "date": "2025-02-20", "categories": []}
{"id": "2502.13399", "title": "MaizeEar-SAM: Zero-Shot Maize Ear Phenotyping", "abstract": "  Quantifying the variation in yield component traits of maize (Zea mays L.),\nwhich together determine the overall productivity of this globally important\ncrop, plays a critical role in plant genetics research, plant breeding, and the\ndevelopment of improved farming practices. Grain yield per acre is calculated\nby multiplying the number of plants per acre, ears per plant, number of kernels\nper ear, and the average kernel weight. The number of kernels per ear is\ndetermined by the number of kernel rows per ear multiplied by the number of\nkernels per row. Traditional manual methods for measuring these two traits are\ntime-consuming, limiting large-scale data collection. Recent automation efforts\nusing image processing and deep learning encounter challenges such as high\nannotation costs and uncertain generalizability.\n  We tackle these issues by exploring Large Vision Models for zero-shot,\nannotation-free maize kernel segmentation. By using an open-source large vision\nmodel, the Segment Anything Model (SAM), we segment individual kernels in RGB\nimages of maize ears and apply a graph-based algorithm to calculate the number\nof kernels per row. Our approach successfully identifies the number of kernels\nper row across a wide range of maize ears, showing the potential of zero-shot\nlearning with foundation vision models combined with image processing\ntechniques to improve automation and reduce subjectivity in agronomic data\ncollection. All our code is open-sourced to make these affordable phenotyping\nmethods accessible to everyone.\n", "date": "2025-02-19", "categories": []}
{"id": "2502.13103", "title": "WeedsGalore: A Multispectral and Multitemporal UAV-based Dataset for\n  Crop and Weed Segmentation in Agricultural Maize Fields", "abstract": "  Weeds are one of the major reasons for crop yield loss but current weeding\npractices fail to manage weeds in an efficient and targeted manner. Effective\nweed management is especially important for crops with high worldwide\nproduction such as maize, to maximize crop yield for meeting increasing global\ndemands. Advances in near-sensing and computer vision enable the development of\nnew tools for weed management. Specifically, state-of-the-art segmentation\nmodels, coupled with novel sensing technologies, can facilitate timely and\naccurate weeding and monitoring systems. However, learning-based approaches\nrequire annotated data and show a lack of generalization to aerial imaging for\ndifferent crops. We present a novel dataset for semantic and instance\nsegmentation of crops and weeds in agricultural maize fields. The multispectral\nUAV-based dataset contains images with RGB, red-edge, and near-infrared bands,\na large number of plant instances, dense annotations for maize and four weed\nclasses, and is multitemporal. We provide extensive baseline results for both\ntasks, including probabilistic methods to quantify prediction uncertainty,\nimprove model calibration, and demonstrate the approach's applicability to\nout-of-distribution data. The results show the effectiveness of the two\nadditional bands compared to RGB only, and better performance in our target\ndomain than models trained on existing datasets. We hope our dataset advances\nresearch on methods and operational systems for fine-grained weed\nidentification, enhancing the robustness and applicability of UAV-based weed\nmanagement. The dataset and code are available at\nhttps://github.com/GFZ/weedsgalore\n", "date": "2025-02-18", "categories": []}
{"id": "2502.08233", "title": "Plantation Monitoring Using Drone Images: A Dataset and Performance\n  Review", "abstract": "  Automatic monitoring of tree plantations plays a crucial role in agriculture.\nFlawless monitoring of tree health helps farmers make informed decisions\nregarding their management by taking appropriate action. Use of drone images\nfor automatic plantation monitoring can enhance the accuracy of the monitoring\nprocess, while still being affordable to small farmers in developing countries\nsuch as India. Small, low cost drones equipped with an RGB camera can capture\nhigh-resolution images of agricultural fields, allowing for detailed analysis\nof the well-being of the plantations. Existing methods of automated plantation\nmonitoring are mostly based on satellite images, which are difficult to get for\nthe farmers. We propose an automated system for plantation health monitoring\nusing drone images, which are becoming easier to get for the farmers. We\npropose a dataset of images of trees with three categories: ``Good health\",\n``Stunted\", and ``Dead\". We annotate the dataset using CVAT annotation tool,\nfor use in research purposes. We experiment with different well-known CNN\nmodels to observe their performance on the proposed dataset. The initial low\naccuracy levels show the complexity of the proposed dataset. Further, our study\nrevealed that, depth-wise convolution operation embedded in a deep CNN model,\ncan enhance the performance of the model on drone dataset. Further, we apply\nstate-of-the-art object detection models to identify individual trees to better\nmonitor them automatically.\n", "date": "2025-02-12", "categories": []}
{"id": "2502.06255", "title": "Towards Efficient and Intelligent Laser Weeding: Method and Dataset for\n  Weed Stem Detection", "abstract": "  Weed control is a critical challenge in modern agriculture, as weeds compete\nwith crops for essential nutrient resources, significantly reducing crop yield\nand quality. Traditional weed control methods, including chemical and\nmechanical approaches, have real-life limitations such as associated\nenvironmental impact and efficiency. An emerging yet effective approach is\nlaser weeding, which uses a laser beam as the stem cutter. Although there have\nbeen studies that use deep learning in weed recognition, its application in\nintelligent laser weeding still requires a comprehensive understanding. Thus,\nthis study represents the first empirical investigation of weed recognition for\nlaser weeding. To increase the efficiency of laser beam cut and avoid damaging\nthe crops of interest, the laser beam shall be directly aimed at the weed root.\nYet, weed stem detection remains an under-explored problem. We integrate the\ndetection of crop and weed with the localization of weed stem into one\nend-to-end system. To train and validate the proposed system in a real-life\nscenario, we curate and construct a high-quality weed stem detection dataset\nwith human annotations. The dataset consists of 7,161 high-resolution pictures\ncollected in the field with annotations of 11,151 instances of weed.\nExperimental results show that the proposed system improves weeding accuracy by\n6.7% and reduces energy cost by 32.3% compared to existing weed recognition\nsystems.\n", "date": "2025-02-10", "categories": []}
{"id": "2501.18835", "title": "Early Diagnosis and Severity Assessment of Weligama Coconut Leaf Wilt\n  Disease and Coconut Caterpillar Infestation using Deep Learning-based Image\n  Processing Techniques", "abstract": "  Global Coconut (Cocos nucifera (L.)) cultivation faces significant\nchallenges, including yield loss, due to pest and disease outbreaks. In\nparticular, Weligama Coconut Leaf Wilt Disease (WCWLD) and Coconut Caterpillar\nInfestation (CCI) damage coconut trees, causing severe coconut production loss\nin Sri Lanka and nearby coconut-producing countries. Currently, both WCWLD and\nCCI are detected through on-field human observations, a process that is not\nonly time-consuming but also limits the early detection of infections. This\npaper presents a study conducted in Sri Lanka, demonstrating the effectiveness\nof employing transfer learning-based Convolutional Neural Network (CNN) and\nMask Region-based-CNN (Mask R-CNN) to identify WCWLD and CCI at their early\nstages and to assess disease progression. Further, this paper presents the use\nof the You Only Look Once (YOLO) object detection model to count the number of\ncaterpillars distributed on leaves with CCI. The introduced methods were tested\nand validated using datasets collected from Matara, Puttalam, and Makandura,\nSri Lanka. The results show that the proposed methods identify WCWLD and CCI\nwith an accuracy of 90% and 95%, respectively. In addition, the proposed WCWLD\ndisease severity identification method classifies the severity with an accuracy\nof 97%. Furthermore, the accuracies of the object detection models for\ncalculating the number of caterpillars in the leaflets were: YOLOv5-96.87%,\nYOLOv8-96.1%, and YOLO11-95.9%.\n", "date": "2025-01-31", "categories": []}
{"id": "2412.13490", "title": "Comparative Analysis of YOLOv9, YOLOv10 and RT-DETR for Real-Time Weed\n  Detection", "abstract": "  This paper presents a comprehensive evaluation of state-of-the-art object\ndetection models, including YOLOv9, YOLOv10, and RT-DETR, for the task of weed\ndetection in smart-spraying applications focusing on three classes: Sugarbeet,\nMonocot, and Dicot. The performance of these models is compared based on mean\nAverage Precision (mAP) scores and inference times on different GPU and CPU\ndevices. We consider various model variations, such as nano, small, medium,\nlarge alongside different image resolutions (320px, 480px, 640px, 800px,\n960px). The results highlight the trade-offs between inference time and\ndetection accuracy, providing valuable insights for selecting the most suitable\nmodel for real-time weed detection. This study aims to guide the development of\nefficient and effective smart spraying systems, enhancing agricultural\nproductivity through precise weed management.\n", "date": "2024-12-18", "categories": []}
{"id": "2502.00939", "title": "Fruit Fly Classification (Diptera: Tephritidae) in Images, Applying\n  Transfer Learning", "abstract": "  This study develops a transfer learning model for the automated\nclassification of two species of fruit flies, Anastrepha fraterculus and\nCeratitis capitata, in a controlled laboratory environment. The research\naddresses the need to optimize identification and classification, which are\ncurrently performed manually by experts, being affected by human factors and\nfacing time challenges. The methodological process of this study includes the\ncapture of high-quality images using a mobile phone camera and a stereo\nmicroscope, followed by segmentation to reduce size and focus on relevant\nmorphological areas. The images were carefully labeled and preprocessed to\nensure the quality and consistency of the dataset used to train the pre-trained\nconvolutional neural network models VGG16, VGG19, and Inception-v3. The results\nwere evaluated using the F1-score, achieving 82% for VGG16 and VGG19, while\nInception-v3 reached an F1-score of 93%. Inception-v3's reliability was\nverified through model testing in uncontrolled environments, with positive\nresults, complemented by the Grad-CAM technique, demonstrating its ability to\ncapture essential morphological features. These findings indicate that\nInception-v3 is an effective and replicable approach for classifying Anastrepha\nfraterculus and Ceratitis capitata, with potential for implementation in\nautomated monitoring systems.\n", "date": "2025-02-02", "categories": []}
{"id": "2501.17387", "title": "Assessing the Capability of YOLO- and Transformer-based Object Detectors\n  for Real-time Weed Detection", "abstract": "  Spot spraying represents an efficient and sustainable method for reducing the\namount of pesticides, particularly herbicides, used in agricultural fields. To\nachieve this, it is of utmost importance to reliably differentiate between\ncrops and weeds, and even between individual weed species in situ and under\nreal-time conditions. To assess suitability for real-time application,\ndifferent object detection models that are currently state-of-the-art are\ncompared. All available models of YOLOv8, YOLOv9, YOLOv10, and RT-DETR are\ntrained and evaluated with images from a real field situation. The images are\nseparated into two distinct datasets: In the initial data set, each species of\nplants is trained individually; in the subsequent dataset, a distinction is\nmade between monocotyledonous weeds, dicotyledonous weeds, and three chosen\ncrops. The results demonstrate that while all models perform equally well in\nthe metrics evaluated, the YOLOv9 models, particularly the YOLOv9s and YOLOv9e,\nstand out in terms of their strong recall scores (66.58 % and 72.36 %), as well\nas mAP50 (73.52 % and 79.86 %), and mAP50-95 (43.82 % and 47.00 %) in dataset\n2. However, the RT-DETR models, especially RT-DETR-l, excel in precision with\nreaching 82.44 \\% on dataset 1 and 81.46 % in dataset 2, making them\nparticularly suitable for scenarios where minimizing false positives is\ncritical. In particular, the smallest variants of the YOLO models (YOLOv8n,\nYOLOv9t, and YOLOv10n) achieve substantially faster inference times down to\n7.58 ms for dataset 2 on the NVIDIA GeForce RTX 4090 GPU for analyzing one\nframe, while maintaining competitive accuracy, highlighting their potential for\ndeployment in resource-constrained embedded computing devices as typically used\nin productive setups.\n", "date": "2025-01-29", "categories": []}
{"id": "2501.16700", "title": "Determining Mosaic Resilience in Sugarcane Plants using Hyperspectral\n  Images", "abstract": "  Sugarcane mosaic disease poses a serious threat to the Australian sugarcane\nindustry, leading to yield losses of up to 30% in susceptible varieties.\nExisting manual inspection methods for detecting mosaic resilience are\ninefficient and impractical for large-scale application. This study introduces\na novel approach using hyperspectral imaging and machine learning to detect\nmosaic resilience by leveraging global feature representation from local\nspectral patches. Hyperspectral data were collected from eight sugarcane\nvarieties under controlled and field conditions. Local spectral patches were\nanalyzed to capture spatial and spectral variations, which were then aggregated\ninto global feature representations using a ResNet18 deep learning\narchitecture. While classical methods like Support Vector Machines struggled to\nutilize spatial-spectral relationships effectively, the deep learning model\nachieved high classification accuracy, demonstrating its capacity to identify\nmosaic resilience from fine-grained hyperspectral data. This approach enhances\nearly detection capabilities, enabling more efficient management of susceptible\nstrains and contributing to sustainable sugarcane production.\n", "date": "2025-01-28", "categories": []}
{"id": "2301.07519", "title": "Site-specific weed management in corn using UAS imagery analysis and\n  computer vision techniques", "abstract": "  Currently, weed control in commercial corn production is performed without\nconsidering weed distribution information in the field. This kind of weed\nmanagement practice leads to excessive amounts of chemical herbicides being\napplied in a given field. The objective of this study was to perform\nsite-specific weed control (SSWC) in a corn field by 1) using an unmanned\naerial system (UAS) to map the spatial distribution information of weeds in the\nfield; 2) creating a prescription map based on the weed distribution map, and\n3) spraying the field using the prescription map and a commercial size sprayer.\nIn this study, we are proposing a Crop Row Identification (CRI) algorithm, a\ncomputer vision algorithm that identifies corn rows on UAS imagery. After being\nidentified, the corn rows were then removed from the imagery and the remaining\nvegetation fraction was classified as weeds. Based on that information, a\ngrid-based weed prescription map was created and the weed control application\nwas implemented through a commercial-size sprayer. The decision of spraying\nherbicides on a particular grid was based on the presence of weeds in that grid\ncell. All the grids that contained at least one weed were sprayed, while the\ngrids free of weeds were not. Using our SSWC approach, we were able to save\n26.23\\% of the land (1.97 acres) from being sprayed with chemical herbicides\ncompared to the existing method. This study presents a full workflow from UAS\nimage collection to field weed control implementation using a commercial-size\nsprayer, and it shows that some level of savings can potentially be obtained\neven in a situation with high weed infestation, which might provide an\nopportunity to reduce chemical usage in corn production systems.\n", "date": "2022-12-31", "categories": []}
{"id": "2410.19869", "title": "Comparing YOLOv11 and YOLOv8 for instance segmentation of occluded and\n  non-occluded immature green fruits in complex orchard environment", "abstract": "  This study conducted a comprehensive performance evaluation on YOLO11 (or\nYOLOv11) and YOLOv8, the latest in the \"You Only Look Once\" (YOLO) series,\nfocusing on their instance segmentation capabilities for immature green apples\nin orchard environments. YOLO11n-seg achieved the highest mask precision across\nall categories with a notable score of 0.831, highlighting its effectiveness in\nfruit detection. YOLO11m-seg and YOLO11l-seg excelled in non-occluded and\noccluded fruitlet segmentation with scores of 0.851 and 0.829, respectively.\nAdditionally, YOLOv11x-seg led in mask recall for all categories, achieving a\nscore of 0.815, with YOLO11m-seg performing best for non-occluded immature\ngreen fruitlets at 0.858 and YOLOv8x-seg leading the occluded category with\n0.800. In terms of mean average precision at a 50\\% intersection over union\n(mAP@50), YOLOv11m-seg consistently outperformed, registering the highest\nscores for both box and mask segmentation, at 0.876 and 0.860 for the \"All\"\nclass and 0.908 and 0.909 for non-occluded immature fruitlets, respectively.\nYOLO11l-seg and YOLOv8l-seg shared the top box mAP@50 for occluded immature\nfruitlets at 0.847, while YOLO11m-seg achieved the highest mask mAP@50 of\n0.810. Despite the advancements in YOLO11, YOLOv8n surpassed its counterparts\nin image processing speed, with an impressive inference speed of 3.3\nmilliseconds, compared to the fastest YOLO11 series model at 4.8 milliseconds,\nunderscoring its suitability for real-time agricultural applications related to\ncomplex green fruit environments. (YOLOv11 segmentation)\n", "date": "2024-10-24", "categories": []}
{"id": "2501.12487", "title": "fabSAM: A Farmland Boundary Delineation Method Based on the Segment\n  Anything Model", "abstract": "  Delineating farmland boundaries is essential for agricultural management such\nas crop monitoring and agricultural census. Traditional methods using remote\nsensing imagery have been efficient but limited in generalisation. The Segment\nAnything Model (SAM), known for its impressive zero shot performance, has been\nadapted for remote sensing tasks through prompt learning and fine tuning. Here,\nwe propose a SAM based farmland boundary delineation framework 'fabSAM' that\ncombines a Deeplabv3+ based Prompter and SAM. Also, a fine tuning strategy was\nintroduced to enable SAMs decoder to improve the use of prompt information.\nExperimental results on the AI4Boundaries and AI4SmallFarms datasets have shown\nthat fabSAM has a significant improvement in farmland region identification and\nboundary delineation. Compared to zero shot SAM, fabSAM surpassed it by 23.5%\nand 15.1% in mIOU on the AI4Boundaries and AI4SmallFarms datasets,\nrespectively. For Deeplabv3+, fabSAM outperformed it by 4.9% and 12.5% in mIOU,\nrespectively. These results highlight the effectiveness of fabSAM, which also\nmeans that we can more easily obtain the global farmland region and boundary\nmaps from open source satellite image datasets like Sentinel2.\n", "date": "2025-01-21", "categories": []}
{"id": "2409.17400", "title": "AgRegNet: A Deep Regression Network for Flower and Fruit Density\n  Estimation, Localization, and Counting in Orchards", "abstract": "  One of the major challenges for the agricultural industry today is the\nuncertainty in manual labor availability and the associated cost. Automated\nflower and fruit density estimation, localization, and counting could help\nstreamline harvesting, yield estimation, and crop-load management strategies\nsuch as flower and fruitlet thinning. This article proposes a deep\nregression-based network, AgRegNet, to estimate density, count, and location of\nflower and fruit in tree fruit canopies without explicit object detection or\npolygon annotation. Inspired by popular U-Net architecture, AgRegNet is a\nU-shaped network with an encoder-to-decoder skip connection and modified\nConvNeXt-T as an encoder feature extractor. AgRegNet can be trained based on\ninformation from point annotation and leverages segmentation information and\nattention modules (spatial and channel) to highlight relevant flower and fruit\nfeatures while suppressing non-relevant background features. Experimental\nevaluation in apple flower and fruit canopy images under an unstructured\norchard environment showed that AgRegNet achieved promising accuracy as\nmeasured by Structural Similarity Index (SSIM), percentage Mean Absolute Error\n(pMAE) and mean Average Precision (mAP) to estimate flower and fruit density,\ncount, and centroid location, respectively. Specifically, the SSIM, pMAE, and\nmAP values for flower images were 0.938, 13.7%, and 0.81, respectively. For\nfruit images, the corresponding values were 0.910, 5.6%, and 0.93. Since the\nproposed approach relies on information from point annotation, it is suitable\nfor sparsely and densely located objects. This simplified technique will be\nhighly applicable for growers to accurately estimate yields and decide on\noptimal chemical and mechanical flower thinning practices.\n", "date": "2024-09-25", "categories": []}
{"id": "2501.10199", "title": "Adaptive Clustering for Efficient Phenotype Segmentation of UAV\n  Hyperspectral Data", "abstract": "  Unmanned Aerial Vehicles (UAVs) combined with Hyperspectral imaging (HSI)\noffer potential for environmental and agricultural applications by capturing\ndetailed spectral information that enables the prediction of invisible features\nlike biochemical leaf properties. However, the data-intensive nature of HSI\nposes challenges for remote devices, which have limited computational resources\nand storage. This paper introduces an Online Hyperspectral Simple Linear\nIterative Clustering algorithm (OHSLIC) framework for real-time tree phenotype\nsegmentation. OHSLIC reduces inherent noise and computational demands through\nadaptive incremental clustering and a lightweight neural network, which\nphenotypes trees using leaf contents such as chlorophyll, carotenoids, and\nanthocyanins. A hyperspectral dataset is created using a custom simulator that\nincorporates realistic leaf parameters, and light interactions. Results\ndemonstrate that OHSLIC achieves superior regression accuracy and segmentation\nperformance compared to pixel- or window-based methods while significantly\nreducing inference time. The method`s adaptive clustering enables dynamic\ntrade-offs between computational efficiency and accuracy, paving the way for\nscalable edge-device deployment in HSI applications.\n", "date": "2025-01-17", "categories": []}
{"id": "2210.01272", "title": "A systematic review of the use of Deep Learning in Satellite Imagery for\n  Agriculture", "abstract": "  Agricultural research is essential for increasing food production to meet the\nrequirements of an increasing population in the coming decades. Recently,\nsatellite technology has been improving rapidly and deep learning has seen much\nsuccess in generic computer vision tasks and many application areas which\npresents an important opportunity to improve analysis of agricultural land.\nHere we present a systematic review of 150 studies to find the current uses of\ndeep learning on satellite imagery for agricultural research. Although we\nidentify 5 categories of agricultural monitoring tasks, the majority of the\nresearch interest is in crop segmentation and yield prediction. We found that,\nwhen used, modern deep learning methods consistently outperformed traditional\nmachine learning across most tasks; the only exception was that Long Short-Term\nMemory (LSTM) Recurrent Neural Networks did not consistently outperform Random\nForests (RF) for yield prediction. The reviewed studies have largely adopted\nmethodologies from generic computer vision, except for one major omission:\nbenchmark datasets are not utilised to evaluate models across studies, making\nit difficult to compare results. Additionally, some studies have specifically\nutilised the extra spectral resolution available in satellite imagery, but\nother divergent properties of satellite images - such as the hugely different\nscales of spatial patterns - are not being taken advantage of in the reviewed\nstudies.\n", "date": "2022-10-03", "categories": []}
{"id": "2404.10073", "title": "Explainable Light-Weight Deep Learning Pipeline for Improved Drought\n  Stress Identification", "abstract": "  Early identification of drought stress in crops is vital for implementing\neffective mitigation measures and reducing yield loss. Non-invasive imaging\ntechniques hold immense potential by capturing subtle physiological changes in\nplants under water deficit. Sensor based imaging data serves as a rich source\nof information for machine learning and deep learning algorithms, facilitating\nfurther analysis aimed at identifying drought stress. While these approaches\nyield favorable results, real-time field applications requires algorithms\nspecifically designed for the complexities of natural agricultural conditions.\nOur work proposes a novel deep learning framework for classifying drought\nstress in potato crops captured by UAVs in natural settings. The novelty lies\nin the synergistic combination of a pre-trained network with carefully designed\ncustom layers. This architecture leverages feature extraction capabilities of\nthe pre-trained network while the custom layers enable targeted dimensionality\nreduction and enhanced regularization, ultimately leading to improved\nperformance. A key innovation of our work involves the integration of\nGradient-Class Activation Mapping (Grad-CAM), an explainability technique.\nGrad-CAM sheds light on the internal workings of the deep learning model,\ntypically referred to as a black box. By visualizing the focus areas of the\nmodel within the images, Grad-CAM fosters interpretability and builds trust in\nthe decision-making process of the model. Our proposed framework achieves\nsuperior performance, particularly with the DenseNet121 pre-trained network,\nreaching a precision of 97% to identify the stressed class with an overall\naccuracy of 91%. Comparative analysis of existing state-of-the-art object\ndetection algorithms reveals the superiority of our approach in significantly\nhigher precision and accuracy.\n", "date": "2024-04-15", "categories": []}
{"id": "2501.02344", "title": "Accurate Crop Yield Estimation of Blueberries using Deep Learning and\n  Smart Drones", "abstract": "  We present an AI pipeline that involves using smart drones equipped with\ncomputer vision to obtain a more accurate fruit count and yield estimation of\nthe number of blueberries in a field. The core components are two\nobject-detection models based on the YOLO deep learning architecture: a Bush\nModel that is able to detect blueberry bushes from images captured at low\naltitudes and at different angles, and a Berry Model that can detect individual\nberries that are visible on a bush. Together, both models allow for more\naccurate crop yield estimation by allowing intelligent control of the drone's\nposition and camera to safely capture side-view images of bushes up close. In\naddition to providing experimental results for our models, which show good\naccuracy in terms of precision and recall when captured images are cropped\naround the foreground center bush, we also describe how to deploy our models to\nmap out blueberry fields using different sampling strategies, and discuss the\nchallenges of annotating very small objects (blueberries) and difficulties in\nevaluating the effectiveness of our models.\n", "date": "2025-01-04", "categories": []}
{"id": "2501.00527", "title": "Exploiting Boundary Loss for the Hierarchical Panoptic Segmentation of\n  Plants and Leaves", "abstract": "  Precision agriculture leverages data and machine learning so that farmers can\nmonitor their crops and target interventions precisely. This enables the\nprecision application of herbicide only to weeds, or the precision application\nof fertilizer only to undernourished crops, rather than to the entire field.\nThe approach promises to maximize yields while minimizing resource use and harm\nto the surrounding environment. To this end, we propose a hierarchical panoptic\nsegmentation method that simultaneously determines leaf count (as an identifier\nof plant growth)and locates weeds within an image. In particular, our approach\naims to improve the segmentation of smaller instances like the leaves and weeds\nby incorporating focal loss and boundary loss. Not only does this result in\ncompetitive performance, achieving a PQ+ of 81.89 on the standard training set,\nbut we also demonstrate we can improve leaf-counting accuracy with our method.\nThe code is available at\nhttps://github.com/madeleinedarbyshire/HierarchicalMask2Former.\n", "date": "2024-12-31", "categories": []}
{"id": "2501.00734", "title": "DDD: Discriminative Difficulty Distance for plant disease diagnosis", "abstract": "  Recent studies on plant disease diagnosis using machine learning (ML) have\nhighlighted concerns about the overestimated diagnostic performance due to\ninappropriate data partitioning, where training and test datasets are derived\nfrom the same source (domain). Plant disease diagnosis presents a challenging\nclassification task, characterized by its fine-grained nature, vague symptoms,\nand the extensive variability of image features within each domain. In this\nstudy, we propose the concept of Discriminative Difficulty Distance (DDD), a\nnovel metric designed to quantify the domain gap between training and test\ndatasets while assessing the classification difficulty of test data. DDD\nprovides a valuable tool for identifying insufficient diversity in training\ndata, thus supporting the development of more diverse and robust datasets. We\ninvestigated multiple image encoders trained on different datasets and examined\nwhether the distances between datasets, measured using low-dimensional\nrepresentations generated by the encoders, are suitable as a DDD metric. The\nstudy utilized 244,063 plant disease images spanning four crops and 34 disease\nclasses collected from 27 domains. As a result, we demonstrated that even if\nthe test images are from different crops or diseases than those used to train\nthe encoder, incorporating them allows the construction of a distance measure\nfor a dataset that strongly correlates with the difficulty of diagnosis\nindicated by the disease classifier developed independently. Compared to the\nbase encoder, pre-trained only on ImageNet21K, the correlation higher by 0.106\nto 0.485, reaching a maximum of 0.909.\n", "date": "2025-01-01", "categories": []}
{"id": "2412.19915", "title": "Identifying Cocoa Pollinators: A Deep Learning Dataset", "abstract": "  Cocoa is a multi-billion-dollar industry but research on improving yields\nthrough pollination remains limited. New embedded hardware and AI-based data\nanalysis is advancing information on cocoa flower visitors, their identity and\nimplications for yields. We present the first cocoa flower visitor dataset\ncontaining 5,792 images of Ceratopogonidae, Formicidae, Aphididae, Araneae, and\nEncyrtidae, and 1,082 background cocoa flower images. This dataset was curated\nfrom 23 million images collected over two years by embedded cameras in cocoa\nplantations in Hainan province, China. We exemplify the use of the dataset with\ndifferent sizes of YOLOv8 models and by progressively increasing the background\nimage ratio in the training set to identify the best-performing model. The\nmedium-sized YOLOv8 model achieved the best results with 8% background images\n(F1 Score of 0.71, mAP50 of 0.70). Overall, this dataset is useful to compare\nthe performance of deep learning model architectures on images with low\ncontrast images and difficult detection targets. The data can support future\nefforts to advance sustainable cocoa production through pollination monitoring\nprojects.\n", "date": "2024-12-27", "categories": []}
{"id": "2412.18635", "title": "Edge-AI for Agriculture: Lightweight Vision Models for Disease Detection\n  in Resource-Limited Settings", "abstract": "  This research paper presents the development of a lightweight and efficient\ncomputer vision pipeline aimed at assisting farmers in detecting orange\ndiseases using minimal resources. The proposed system integrates advanced\nobject detection, classification, and segmentation models, optimized for\ndeployment on edge devices, ensuring functionality in resource-limited\nenvironments. The study evaluates the performance of various state-of-the-art\nmodels, focusing on their accuracy, computational efficiency, and\ngeneralization capabilities. Notable findings include the Vision Transformer\nachieving 96 accuracy in orange species classification and the lightweight\nYOLOv8-S model demonstrating exceptional object detection performance with\nminimal computational overhead. The research highlights the potential of modern\ndeep learning architectures to address critical agricultural challenges,\nemphasizing the importance of model complexity versus practical utility. Future\nwork will explore expanding datasets, model compression techniques, and\nfederated learning to enhance the applicability of these systems in diverse\nagricultural contexts, ultimately contributing to more sustainable farming\npractices.\n", "date": "2024-12-23", "categories": []}
{"id": "2412.17142", "title": "AI-Based Teat Shape and Skin Condition Prediction for Dairy Management", "abstract": "  Dairy owners spend significant effort to keep their animals healthy. There is\ngood reason to hope that technologies such as computer vision and artificial\nintelligence (AI) could reduce these costs, yet obstacles arise when adapting\nadvanced tools to farming environments. In this work, we adapt AI tools to\ndairy cow teat localization, teat shape, and teat skin condition\nclassifications. We also curate a data collection and analysis methodology for\na Machine Learning (ML) pipeline. The resulting teat shape prediction model\nachieves a mean Average Precision (mAP) of 0.783, and the teat skin condition\nmodel achieves a mean average precision of 0.828. Our work leverages existing\nML vision models to facilitate the individualized identification of teat health\nand skin conditions, applying AI to the dairy management industry.\n", "date": "2024-12-22", "categories": []}
{"id": "2409.16252", "title": "Fields of The World: A Machine Learning Benchmark Dataset For Global\n  Agricultural Field Boundary Segmentation", "abstract": "  Crop field boundaries are foundational datasets for agricultural monitoring\nand assessments but are expensive to collect manually. Machine learning (ML)\nmethods for automatically extracting field boundaries from remotely sensed\nimages could help realize the demand for these datasets at a global scale.\nHowever, current ML methods for field instance segmentation lack sufficient\ngeographic coverage, accuracy, and generalization capabilities. Further,\nresearch on improving ML methods is restricted by the lack of labeled datasets\nrepresenting the diversity of global agricultural fields. We present Fields of\nThe World (FTW) -- a novel ML benchmark dataset for agricultural field instance\nsegmentation spanning 24 countries on four continents (Europe, Africa, Asia,\nand South America). FTW is an order of magnitude larger than previous datasets\nwith 70,462 samples, each containing instance and semantic segmentation masks\npaired with multi-date, multi-spectral Sentinel-2 satellite images. We provide\nresults from baseline models for the new FTW benchmark, show that models\ntrained on FTW have better zero-shot and fine-tuning performance in held-out\ncountries than models that aren't pre-trained with diverse datasets, and show\npositive qualitative zero-shot results of FTW models in a real-world scenario\n-- running on Sentinel-2 scenes over Ethiopia.\n", "date": "2024-09-24", "categories": []}
{"id": "2412.14954", "title": "Corn Ear Detection and Orientation Estimation Using Deep Learning", "abstract": "  Monitoring growth behavior of maize plants such as the development of ears\ncan give key insights into the plant's health and development. Traditionally,\nthe measurement of the angle of ears is performed manually, which can be\ntime-consuming and prone to human error. To address these challenges, this\npaper presents a computer vision-based system for detecting and tracking ears\nof corn in an image sequence. The proposed system could accurately detect,\ntrack, and predict the ear's orientation, which can be useful in monitoring\ntheir growth behavior. This can significantly save time compared to manual\nmeasurement and enables additional areas of ear orientation research and\npotential increase in efficiencies for maize production. Using an object\ndetector with keypoint detection, the algorithm proposed could detect 90\npercent of all ears. The cardinal estimation had a mean absolute error (MAE) of\n18 degrees, compared to a mean 15 degree difference between two people\nmeasuring by hand. These results demonstrate the feasibility of using computer\nvision techniques for monitoring maize growth and can lead to further research\nin this area.\n", "date": "2024-12-19", "categories": []}
{"id": "2412.12949", "title": "Synthetic Data Generation for Anomaly Detection on Table Grapes", "abstract": "  Early detection of illnesses and pest infestations in fruit cultivation is\ncritical for maintaining yield quality and plant health. Computer vision and\nrobotics are increasingly employed for the automatic detection of such issues,\nparticularly using data-driven solutions. However, the rarity of these problems\nmakes acquiring and processing the necessary data to train such algorithms a\nsignificant obstacle. One solution to this scarcity is the generation of\nsynthetic high-quality anomalous samples. While numerous methods exist for this\ntask, most require highly trained individuals for setup.\n  This work addresses the challenge of generating synthetic anomalies in an\nautomatic fashion that requires only an initial collection of normal and\nanomalous samples from the user - a task that is straightforward for farmers.\nWe demonstrate the approach in the context of table grape cultivation.\nSpecifically, based on the observation that normal berries present relatively\nsmooth surfaces, while defects result in more complex textures, we introduce a\nDual-Canny Edge Detection (DCED) filter. This filter emphasizes the additional\ntexture indicative of diseases, pest infestations, or other defects. Using\nsegmentation masks provided by the Segment Anything Model, we then select and\nseamlessly blend anomalous berries onto normal ones. We show that the proposed\ndataset augmentation technique improves the accuracy of an anomaly classifier\nfor table grapes and that the approach can be generalized to other fruit types.\n", "date": "2024-12-17", "categories": []}
{"id": "2412.12966", "title": "Fruit Deformity Classification through Single-Input and Multi-Input\n  Architectures based on CNN Models using Real and Synthetic Images", "abstract": "  The present study focuses on detecting the degree of deformity in fruits such\nas apples, mangoes, and strawberries during the process of inspecting their\nexternal quality, employing Single-Input and Multi-Input architectures based on\nconvolutional neural network (CNN) models using sets of real and synthetic\nimages. The datasets are segmented using the Segment Anything Model (SAM),\nwhich provides the silhouette of the fruits. Regarding the single-input\narchitecture, the evaluation of the CNN models is performed only with real\nimages, but a methodology is proposed to improve these results using a\npre-trained model with synthetic images. In the Multi-Input architecture,\nbranches with RGB images and fruit silhouettes are implemented as inputs for\nevaluating CNN models such as VGG16, MobileNetV2, and CIDIS. However, the\nresults revealed that the Multi-Input architecture with the MobileNetV2 model\nwas the most effective in identifying deformities in the fruits, achieving\naccuracies of 90\\%, 94\\%, and 92\\% for apples, mangoes, and strawberries,\nrespectively. In conclusion, the Multi-Input architecture with the MobileNetV2\nmodel is the most accurate for classifying levels of deformity in fruits.\n", "date": "2024-12-17", "categories": []}
{"id": "2412.10464", "title": "Automatic Detection, Positioning and Counting of Grape Bunches Using\n  Robots", "abstract": "  In order to promote agricultural automatic picking and yield estimation\ntechnology, this project designs a set of automatic detection, positioning and\ncounting algorithms for grape bunches, and applies it to agricultural robots.\nThe Yolov3 detection network is used to realize the accurate detection of grape\nbunches, and the local tracking algorithm is added to eliminate relocation.\nThen it obtains the accurate 3D spatial position of the central points of grape\nbunches using the depth distance and the spatial restriction method. Finally,\nthe counting of grape bunches is completed. It is verified using the\nagricultural robot in the simulated vineyard environment. The project code is\nreleased at:\nhttps://github.com/XuminGaoGithub/Grape_bunches_count_using_robots.\n", "date": "2024-12-12", "categories": []}
{"id": "2412.11949", "title": "Coconut Palm Tree Counting on Drone Images with Deep Object Detection\n  and Synthetic Training Data", "abstract": "  Drones have revolutionized various domains, including agriculture. Recent\nadvances in deep learning have propelled among other things object detection in\ncomputer vision. This study utilized YOLO, a real-time object detector, to\nidentify and count coconut palm trees in Ghanaian farm drone footage. The farm\npresented has lost track of its trees due to different planting phases. While\nmanual counting would be very tedious and error-prone, accurately determining\nthe number of trees is crucial for efficient planning and management of\nagricultural processes, especially for optimizing yields and predicting\nproduction. We assessed YOLO for palm detection within a semi-automated\nframework, evaluated accuracy augmentations, and pondered its potential for\nfarmers. Data was captured in September 2022 via drones. To optimize YOLO with\nscarce data, synthetic images were created for model training and validation.\nThe YOLOv7 model, pretrained on the COCO dataset (excluding coconut palms), was\nadapted using tailored data. Trees from footage were repositioned on synthetic\nimages, with testing on distinct authentic images. In our experiments, we\nadjusted hyperparameters, improving YOLO's mean average precision (mAP). We\nalso tested various altitudes to determine the best drone height. From an\ninitial mAP@.5 of $0.65$, we achieved 0.88, highlighting the value of synthetic\nimages in agricultural scenarios.\n", "date": "2024-12-16", "categories": []}
{"id": "2412.09696", "title": "Soybean Maturity Prediction using 2D Contour Plots from Drone based Time\n  Series Imagery", "abstract": "  Plant breeding programs require assessments of days to maturity for accurate\nselection and placement of entries in appropriate tests. In the early stages of\nthe breeding pipeline, soybean breeding programs assign relative maturity\nratings to experimental varieties that indicate their suitable maturity zones.\nTraditionally, the estimation of maturity value for breeding varieties has\ninvolved breeders manually inspecting fields and assessing maturity value\nvisually. This approach relies heavily on rater judgment, making it subjective\nand time-consuming. This study aimed to develop a machine-learning model for\nevaluating soybean maturity using UAV-based time-series imagery. Images were\ncaptured at three-day intervals, beginning as the earliest varieties started\nmaturing and continuing until the last varieties fully matured. The data\ncollected for this experiment consisted of 22,043 plots collected across three\nyears (2021 to 2023) and represent relative maturity groups 1.6 - 3.9. We\nutilized contour plot images extracted from the time-series UAV RGB imagery as\ninput for a neural network model. This contour plot approach encoded the\ntemporal and spatial variation within each plot into a single image. A deep\nlearning model was trained to utilize this contour plot to predict maturity\nratings. This model significantly improves accuracy and robustness, achieving\nup to 85% accuracy. We also evaluate the model's accuracy as we reduce the\nnumber of time points, quantifying the trade-off between temporal resolution\nand maturity prediction. The predictive model offers a scalable, objective, and\nefficient means of assessing crop maturity, enabling phenomics and ML\napproaches to reduce the reliance on manual inspection and subjective\nassessment. This approach enables the automatic prediction of relative maturity\nratings in a breeding program, saving time and resources.\n", "date": "2024-12-12", "categories": []}
{"id": "2412.09739", "title": "Agtech Framework for Cranberry-Ripening Analysis Using Vision Foundation\n  Models", "abstract": "  Agricultural domains are being transformed by recent advances in AI and\ncomputer vision that support quantitative visual evaluation. Using aerial and\nground imaging over a time series, we develop a framework for characterizing\nthe ripening process of cranberry crops, a crucial component for precision\nagriculture tasks such as comparing crop breeds (high-throughput phenotyping)\nand detecting disease. Using drone imaging, we capture images from 20 waypoints\nacross multiple bogs, and using ground-based imaging (hand-held camera), we\nimage same bog patch using fixed fiducial markers. Both imaging methods are\nrepeated to gather a multi-week time series spanning the entire growing season.\nAerial imaging provides multiple samples to compute a distribution of albedo\nvalues. Ground imaging enables tracking of individual berries for a detailed\nview of berry appearance changes. Using vision transformers (ViT) for feature\ndetection after segmentation, we extract a high dimensional feature descriptor\nof berry appearance. Interpretability of appearance is critical for plant\nbiologists and cranberry growers to support crop breeding decisions (e.g.\\\ncomparison of berry varieties from breeding programs). For interpretability, we\ncreate a 2D manifold of cranberry appearance by using a UMAP dimensionality\nreduction on ViT features. This projection enables quantification of ripening\npaths and a useful metric of ripening rate. We demonstrate the comparison of\nfour cranberry varieties based on our ripening assessments. This work is the\nfirst of its kind and has future impact for cranberries and for other crops\nincluding wine grapes, olives, blueberries, and maize. Aerial and ground\ndatasets are made publicly available.\n", "date": "2024-12-12", "categories": []}
{"id": "2412.07408", "title": "Explainability of Deep Learning-Based Plant Disease Classifiers Through\n  Automated Concept Identification", "abstract": "  While deep learning has significantly advanced automatic plant disease\ndetection through image-based classification, improving model explainability\nremains crucial for reliable disease detection. In this study, we apply the\nAutomated Concept-based Explanation (ACE) method to plant disease\nclassification using the widely adopted InceptionV3 model and the PlantVillage\ndataset. ACE automatically identifies the visual concepts found in the image\ndata and provides insights about the critical features influencing the model\npredictions. This approach reveals both effective disease-related patterns and\nincidental biases, such as those from background or lighting that can\ncompromise model robustness. Through systematic experiments, ACE helped us to\nidentify relevant features and pinpoint areas for targeted model improvement.\nOur findings demonstrate the potential of ACE to improve the explainability of\nplant disease classification based on deep learning, which is essential for\nproducing transparent tools for plant disease management in agriculture.\n", "date": "2024-12-10", "categories": []}
{"id": "2412.05728", "title": "Integrating YOLO11 and Convolution Block Attention Module for\n  Multi-Season Segmentation of Tree Trunks and Branches in Commercial Apple\n  Orchards", "abstract": "  In this study, we developed a customized instance segmentation model by\nintegrating the Convolutional Block Attention Module (CBAM) with the YOLO11\narchitecture. This model, trained on a mixed dataset of dormant and canopy\nseason apple orchard images, aimed to enhance the segmentation of tree trunks\nand branches under varying seasonal conditions throughout the year. The model\nwas individually validated across dormant and canopy season images after\ntraining the YOLO11-CBAM on the mixed dataset collected over the two seasons.\nAdditional testing of the model during pre-bloom, flower bloom, fruit thinning,\nand harvest season was performed. The highest recall and precision metrics were\nobserved in the YOLO11x-seg-CBAM and YOLO11m-seg-CBAM respectively.\nParticularly, YOLO11m-seg with CBAM showed the highest precision of 0.83 as\nperformed for the Trunk class in training, while without the CBAM, YOLO11m-seg\nachieved 0.80 precision score for the Trunk class. Likewise, for branch class,\nYOLO11m-seg with CBAM achieved the highest precision score value of 0.75 while\nwithout the CBAM, the YOLO11m-seg achieved a precision of 0.73. For dormant\nseason validation, YOLO11x-seg exhibited the highest precision at 0.91. Canopy\nseason validation highlighted YOLO11s-seg with superior precision across all\nclasses, achieving 0.516 for Branch, and 0.64 for Trunk. The modeling approach,\ntrained on two season datasets as dormant and canopy season images,\ndemonstrated the potential of the YOLO11-CBAM integration to effectively detect\nand segment tree trunks and branches year-round across all seasonal variations.\nKeywords: YOLOv11, YOLOv11 Tree Detection, YOLOv11 Branch Detection and\nSegmentation, Machine Vision, Deep Learning, Machine Learning\n", "date": "2024-12-07", "categories": []}
{"id": "2408.11253", "title": "Automatic Image Annotation (AIA) of AlmondNet-20 Method for Almond\n  Detection by Improved CNN-based Model", "abstract": "  In response to the burgeoning global demand for premium agricultural\nproducts, particularly within the competitive nut market, this paper introduces\nan innovative methodology aimed at enhancing the grading process for almonds\nand their shells. Leveraging state-of-the-art Deep Convolutional Neural\nNetworks (CNNs), specifically the AlmondNet-20 architecture, our study achieves\nexceptional accuracy exceeding 99%, facilitated by the utilization of a\n20-layer CNN model. To bolster robustness in differentiating between almonds\nand shells, data augmentation techniques are employed, ensuring the reliability\nand accuracy of our classification system. Our model, meticulously trained over\n1000 epochs, demonstrates remarkable performance, boasting an accuracy rate of\n99% alongside a minimal loss function of 0.0567. Rigorous evaluation through\ntest datasets further validates the efficacy of our approach, revealing\nimpeccable precision, recall, and F1-score metrics for almond detection. Beyond\nits technical prowess, this advanced classification system offers tangible\nbenefits to both industry experts and non-specialists alike, ensuring globally\nreliable almond classification. The application of deep learning algorithms, as\nshowcased in our study, not only enhances grading accuracy but also presents\nopportunities for product patents, thereby contributing to the economic value\nof our nation. Through the adoption of cutting-edge technologies such as the\nAlmondNet-20 model, we pave the way for future advancements in agricultural\nproduct classification, ultimately enriching global trade and economic\nprosperity.\n", "date": "2024-08-21", "categories": []}
{"id": "2412.02642", "title": "Robust soybean seed yield estimation using high-throughput ground robot\n  videos", "abstract": "  We present a novel method for soybean (Glycine max (L.) Merr.) yield\nestimation leveraging high throughput seed counting via computer vision and\ndeep learning techniques. Traditional methods for collecting yield data are\nlabor-intensive, costly, prone to equipment failures at critical data\ncollection times, and require transportation of equipment across field sites.\nComputer vision, the field of teaching computers to interpret visual data,\nallows us to extract detailed yield information directly from images. By\ntreating it as a computer vision task, we report a more efficient alternative,\nemploying a ground robot equipped with fisheye cameras to capture comprehensive\nvideos of soybean plots from which images are extracted in a variety of\ndevelopment programs. These images are processed through the P2PNet-Yield\nmodel, a deep learning framework where we combined a Feature Extraction Module\n(the backbone of the P2PNet-Soy) and a Yield Regression Module to estimate seed\nyields of soybean plots. Our results are built on three years of yield testing\nplot data - 8500 in 2021, 2275 in 2022, and 650 in 2023. With these datasets,\nour approach incorporates several innovations to further improve the accuracy\nand generalizability of the seed counting and yield estimation architecture,\nsuch as the fisheye image correction and data augmentation with random sensor\neffects. The P2PNet-Yield model achieved a genotype ranking accuracy score of\nup to 83%. It demonstrates up to a 32% reduction in time to collect yield data\nas well as costs associated with traditional yield estimation, offering a\nscalable solution for breeding programs and agricultural productivity\nenhancement.\n", "date": "2024-12-03", "categories": []}
{"id": "2407.21341", "title": "High-throughput 3D shape completion of potato tubers on a harvester", "abstract": "  Potato yield is an important metric for farmers to further optimize their\ncultivation practices. Potato yield can be estimated on a harvester using an\nRGB-D camera that can estimate the three-dimensional (3D) volume of individual\npotato tubers. A challenge, however, is that the 3D shape derived from RGB-D\nimages is only partially completed, underestimating the actual volume. To\naddress this issue, we developed a 3D shape completion network, called CoRe++,\nwhich can complete the 3D shape from RGB-D images. CoRe++ is a deep learning\nnetwork that consists of a convolutional encoder and a decoder. The encoder\ncompresses RGB-D images into latent vectors that are used by the decoder to\ncomplete the 3D shape using the deep signed distance field network (DeepSDF).\nTo evaluate our CoRe++ network, we collected partial and complete 3D point\nclouds of 339 potato tubers on an operational harvester in Japan. On the 1425\nRGB-D images in the test set (representing 51 unique potato tubers), our\nnetwork achieved a completion accuracy of 2.8 mm on average. For volumetric\nestimation, the root mean squared error (RMSE) was 22.6 ml, and this was better\nthan the RMSE of the linear regression (31.1 ml) and the base model (36.9 ml).\nWe found that the RMSE can be further reduced to 18.2 ml when performing the 3D\nshape completion in the center of the RGB-D image. With an average 3D shape\ncompletion time of 10 milliseconds per tuber, we can conclude that CoRe++ is\nboth fast and accurate enough to be implemented on an operational harvester for\nhigh-throughput potato yield estimation. CoRe++'s high-throughput and accurate\nprocessing allows it to be applied to other tuber, fruit and vegetable crops,\nthereby enabling versatile, accurate and real-time yield monitoring in\nprecision agriculture. Our code, network weights and dataset are publicly\navailable at https://github.com/UTokyo-FieldPhenomics-Lab/corepp.git.\n", "date": "2024-07-31", "categories": []}
{"id": "2411.18513", "title": "Enhancing weed detection performance by means of GenAI-based image\n  augmentation", "abstract": "  Precise weed management is essential for sustaining crop productivity and\necological balance. Traditional herbicide applications face economic and\nenvironmental challenges, emphasizing the need for intelligent weed control\nsystems powered by deep learning. These systems require vast amounts of\nhigh-quality training data. The reality of scarcity of well-annotated training\ndata, however, is often addressed through generating more data using data\naugmentation. Nevertheless, conventional augmentation techniques such as random\nflipping, color changes, and blurring lack sufficient fidelity and diversity.\nThis paper investigates a generative AI-based augmentation technique that uses\nthe Stable Diffusion model to produce diverse synthetic images that improve the\nquantity and quality of training datasets for weed detection models. Moreover,\nthis paper explores the impact of these synthetic images on the performance of\nreal-time detection systems, thus focusing on compact CNN-based models such as\nYOLO nano for edge devices. The experimental results show substantial\nimprovements in mean Average Precision (mAP50 and mAP50-95) scores for YOLO\nmodels trained with generative AI-augmented datasets, demonstrating the\npromising potential of synthetic data to enhance model robustness and accuracy.\n", "date": "2024-11-27", "categories": []}
{"id": "2411.16989", "title": "CMAViT: Integrating Climate, Managment, and Remote Sensing Data for Crop\n  Yield Estimation with Multimodel Vision Transformers", "abstract": "  Crop yield prediction is essential for agricultural planning but remains\nchallenging due to the complex interactions between weather, climate, and\nmanagement practices. To address these challenges, we introduce a deep\nlearning-based multi-model called Climate-Management Aware Vision Transformer\n(CMAViT), designed for pixel-level vineyard yield predictions. CMAViT\nintegrates both spatial and temporal data by leveraging remote sensing imagery\nand short-term meteorological data, capturing the effects of growing season\nvariations. Additionally, it incorporates management practices, which are\nrepresented in text form, using a cross-attention encoder to model their\ninteraction with time-series data. This innovative multi-modal transformer\ntested on a large dataset from 2016-2019 covering 2,200 hectares and eight\ngrape cultivars including more than 5 million vines, outperforms traditional\nmodels like UNet-ConvLSTM, excelling in spatial variability capture and yield\nprediction, particularly for extreme values in vineyards. CMAViT achieved an R2\nof 0.84 and a MAPE of 8.22% on an unseen test dataset. Masking specific\nmodalities lowered performance: excluding management practices, climate data,\nand both reduced R2 to 0.73, 0.70, and 0.72, respectively, and raised MAPE to\n11.92%, 12.66%, and 12.39%, highlighting each modality's importance for\naccurate yield prediction. Code is available at\nhttps://github.com/plant-ai-biophysics-lab/CMAViT.\n", "date": "2024-11-25", "categories": []}
{"id": "2411.16132", "title": "TreeFormer: Single-view Plant Skeleton Estimation via Tree-constrained\n  Graph Generation", "abstract": "  Accurate estimation of plant skeletal structure (e.g., branching structure)\nfrom images is essential for smart agriculture and plant science. Unlike human\nskeletons with fixed topology, plant skeleton estimation presents a unique\nchallenge, i.e., estimating arbitrary tree graphs from images. While recent\ngraph generation methods successfully infer thin structures from images, it is\nchallenging to constrain the output graph strictly to a tree structure. To this\nproblem, we present TreeFormer, a plant skeleton estimator via tree-constrained\ngraph generation. Our approach combines learning-based graph generation with\ntraditional graph algorithms to impose the constraints during the training\nloop. Specifically, our method projects an unconstrained graph onto a minimum\nspanning tree (MST) during the training loop and incorporates this prior\nknowledge into the gradient descent optimization by suppressing unwanted\nfeature values. Experiments show that our method accurately estimates target\nplant skeletal structures for multiple domains: Synthetic tree patterns, real\nbotanical roots, and grapevine branches. Our implementations are available at\nhttps://github.com/huntorochi/TreeFormer/.\n", "date": "2024-11-25", "categories": []}
{"id": "2403.10722", "title": "Cannabis Seed Variant Detection using Faster R-CNN", "abstract": "  Analyzing and detecting cannabis seed variants is crucial for the agriculture\nindustry. It enables precision breeding, allowing cultivators to selectively\nenhance desirable traits. Accurate identification of seed variants also ensures\nregulatory compliance, facilitating the cultivation of specific cannabis\nstrains with defined characteristics, ultimately improving agricultural\nproductivity and meeting diverse market demands. This paper presents a study on\ncannabis seed variant detection by employing a state-of-the-art object\ndetection model Faster R-CNN. This study implemented the model on a locally\nsourced cannabis seed dataset in Thailand, comprising 17 distinct classes. We\nevaluate six Faster R-CNN models by comparing performance on various metrics\nand achieving a mAP score of 94.08\\% and an F1 score of 95.66\\%. This paper\npresents the first known application of deep neural network object detection\nmodels to the novel task of visually identifying cannabis seed types.\n", "date": "2024-03-15", "categories": []}
{"id": "2411.10357", "title": "Interactive Image-Based Aphid Counting in Yellow Water Traps under\n  Stirring Actions", "abstract": "  The current vision-based aphid counting methods in water traps suffer from\nundercounts caused by occlusions and low visibility arising from dense\naggregation of insects and other objects. To address this problem, we propose a\nnovel aphid counting method through interactive stirring actions. We use\ninteractive stirring to alter the distribution of aphids in the yellow water\ntrap and capture a sequence of images which are then used for aphid detection\nand counting through an optimized small object detection network based on\nYolov5. We also propose a counting confidence evaluation system to evaluate the\nconfidence of count-ing results. The final counting result is a weighted sum of\nthe counting results from all sequence images based on the counting confidence.\nExperimental results show that our proposed aphid detection network\nsignificantly outperforms the original Yolov5, with improvements of 33.9% in\nAP@0.5 and 26.9% in AP@[0.5:0.95] on the aphid test set. In addition, the aphid\ncounting test results using our proposed counting confidence evaluation system\nshow significant improvements over the static counting method, closely aligning\nwith manual counting results.\n", "date": "2024-11-15", "categories": []}
{"id": "2411.08453", "title": "Biomass phenotyping of oilseed rape through UAV multi-view oblique\n  imaging with 3DGS and SAM model", "abstract": "  Biomass estimation of oilseed rape is crucial for optimizing crop\nproductivity and breeding strategies. While UAV-based imaging has advanced\nhigh-throughput phenotyping, current methods often rely on orthophoto images,\nwhich struggle with overlapping leaves and incomplete structural information in\ncomplex field environments. This study integrates 3D Gaussian Splatting (3DGS)\nwith the Segment Anything Model (SAM) for precise 3D reconstruction and biomass\nestimation of oilseed rape. UAV multi-view oblique images from 36 angles were\nused to perform 3D reconstruction, with the SAM module enhancing point cloud\nsegmentation. The segmented point clouds were then converted into point cloud\nvolumes, which were fitted to ground-measured biomass using linear regression.\nThe results showed that 3DGS (7k and 30k iterations) provided high accuracy,\nwith peak signal-to-noise ratios (PSNR) of 27.43 and 29.53 and training times\nof 7 and 49 minutes, respectively. This performance exceeded that of structure\nfrom motion (SfM) and mipmap Neural Radiance Fields (Mip-NeRF), demonstrating\nsuperior efficiency. The SAM module achieved high segmentation accuracy, with a\nmean intersection over union (mIoU) of 0.961 and an F1-score of 0.980.\nAdditionally, a comparison of biomass extraction models found the point cloud\nvolume model to be the most accurate, with an determination coefficient (R2) of\n0.976, root mean square error (RMSE) of 2.92 g/plant, and mean absolute\npercentage error (MAPE) of 6.81%, outperforming both the plot crop volume and\nindividual crop volume models. This study highlights the potential of combining\n3DGS with multi-view UAV imaging for improved biomass phenotyping.\n", "date": "2024-11-13", "categories": []}
{"id": "2411.06702", "title": "Track Any Peppers: Weakly Supervised Sweet Pepper Tracking Using VLMs", "abstract": "  In the Detection and Multi-Object Tracking of Sweet Peppers Challenge, we\npresent Track Any Peppers (TAP) - a weakly supervised ensemble technique for\nsweet peppers tracking. TAP leverages the zero-shot detection capabilities of\nvision-language foundation models like Grounding DINO to automatically generate\npseudo-labels for sweet peppers in video sequences with minimal human\nintervention. These pseudo-labels, refined when necessary, are used to train a\nYOLOv8 segmentation network. To enhance detection accuracy under challenging\nconditions, we incorporate pre-processing techniques such as relighting\nadjustments and apply depth-based filtering during post-inference. For object\ntracking, we integrate the Matching by Segment Anything (MASA) adapter with the\nBoT-SORT algorithm. Our approach achieves a HOTA score of 80.4%, MOTA of 66.1%,\nRecall of 74.0%, and Precision of 90.7%, demonstrating effective tracking of\nsweet peppers without extensive manual effort. This work highlights the\npotential of foundation models for efficient and accurate object detection and\ntracking in agricultural settings.\n", "date": "2024-11-11", "categories": []}
{"id": "2411.02762", "title": "EcoCropsAID: Economic Crops Aerial Image Dataset for Land Use\n  Classification", "abstract": "  The EcoCropsAID dataset is a comprehensive collection of 5,400 aerial images\ncaptured between 2014 and 2018 using the Google Earth application. This dataset\nfocuses on five key economic crops in Thailand: rice, sugarcane, cassava,\nrubber, and longan. The images were collected at various crop growth stages:\nearly cultivation, growth, and harvest, resulting in significant variability\nwithin each category and similarities across different categories. These\nvariations, coupled with differences in resolution, color, and contrast\nintroduced by multiple remote imaging sensors, present substantial challenges\nfor land use classification. The dataset is an interdisciplinary resource that\nspans multiple research domains, including remote sensing, geoinformatics,\nartificial intelligence, and computer vision. The unique features of the\nEcoCropsAID dataset offer opportunities for researchers to explore novel\napproaches, such as extracting spatial and temporal features, developing deep\nlearning architectures, and implementing transformer-based models. The\nEcoCropsAID dataset provides a valuable platform for advancing research in land\nuse classification, with implications for optimizing agricultural practices and\nenhancing sustainable development. This study explicitly investigates the use\nof deep learning algorithms to classify economic crop areas in northeastern\nThailand, utilizing satellite imagery to address the challenges posed by\ndiverse patterns and similarities across categories.\n", "date": "2024-11-05", "categories": []}
{"id": "2411.03114", "title": "Investigating the Applicability of a Snapshot Computed Tomography\n  Imaging Spectrometer for the Prediction of Brix and pH of Grapes", "abstract": "  In this paper, a recently developed snapshot hyperspectral imaging (HSI)\nsystem based on Computed Tomography Imaging Spectroscopy (CTIS) is utilized to\ndetermine Brix and pH values in Sheegene 20 table grapes through Partial Least\nSquares Regression (PLSR) modeling. The performance of the CTIS system is\ncompared with that of a state-of-the-art line scan HSI system by imaging 100\ngrapes across both platforms. Reference measurements of Brix and pH values are\nobtained directly using a refractometer and a pH meter, as these parameters are\nessential for assessing the quality of table and wine grapes. The findings\nindicate that the spectra captured by the CTIS camera correlate well with the\nreference measurements, despite the system's narrower spectral range. The CTIS\ncamera's advantages, including its lower cost, portability, and reduced\nsusceptibility to motion errors, highlight its potential for promising in-field\napplications in grape quality assessment.\n", "date": "2024-11-05", "categories": []}
{"id": "2411.00967", "title": "Raspberry PhenoSet: A Phenology-based Dataset for Automated Growth\n  Detection and Yield Estimation", "abstract": "  The future of the agriculture industry is intertwined with automation.\nAccurate fruit detection, yield estimation, and harvest time estimation are\ncrucial for optimizing agricultural practices. These tasks can be carried out\nby robots to reduce labour costs and improve the efficiency of the process. To\ndo so, deep learning models should be trained to perform knowledge-based tasks,\nwhich outlines the importance of contributing valuable data to the literature.\nIn this paper, we introduce Raspberry PhenoSet, a phenology-based dataset\ndesigned for detecting and segmenting raspberry fruit across seven\ndevelopmental stages. To the best of our knowledge, Raspberry PhenoSet is the\nfirst fruit dataset to integrate biology-based classification with fruit\ndetection tasks, offering valuable insights for yield estimation and precise\nharvest timing. This dataset contains 1,853 high-resolution images, the highest\nquality in the literature, captured under controlled artificial lighting in a\nvertical farm. The dataset has a total of 6,907 instances of mask annotations,\nmanually labelled to reflect the seven phenology stages. We have also\nbenchmarked Raspberry PhenoSet using several state-of-the-art deep learning\nmodels, including YOLOv8, YOLOv10, RT-DETR, and Mask R-CNN, to provide a\ncomprehensive evaluation of their performance on the dataset. Our results\nhighlight the challenges of distinguishing subtle phenology stages and\nunderscore the potential of Raspberry PhenoSet for both deep learning model\ndevelopment and practical robotic applications in agriculture, particularly in\nyield prediction and supply chain management. The dataset and the trained\nmodels are publicly available for future studies.\n", "date": "2024-11-01", "categories": []}
{"id": "2411.00548", "title": "Generative AI-based Pipeline Architecture for Increasing Training\n  Efficiency in Intelligent Weed Control Systems", "abstract": "  In automated crop protection tasks such as weed control, disease diagnosis,\nand pest monitoring, deep learning has demonstrated significant potential.\nHowever, these advanced models rely heavily on high-quality, diverse datasets,\noften limited and costly in agricultural settings. Traditional data\naugmentation can increase dataset volume but usually lacks the real-world\nvariability needed for robust training. This study presents a new approach for\ngenerating synthetic images to improve deep learning-based object detection\nmodels for intelligent weed control. Our GenAI-based image generation pipeline\nintegrates the Segment Anything Model (SAM) for zero-shot domain adaptation\nwith a text-to-image Stable Diffusion Model, enabling the creation of synthetic\nimages that capture diverse real-world conditions. We evaluate these synthetic\ndatasets using lightweight YOLO models, measuring data efficiency with mAP50\nand mAP50-95 scores across varying proportions of real and synthetic data.\nNotably, YOLO models trained on datasets with 10% synthetic and 90% real images\ngenerally demonstrate superior mAP50 and mAP50-95 scores compared to those\ntrained solely on real images. This approach not only reduces dependence on\nextensive real-world datasets but also enhances predictive performance. The\nintegration of this approach opens opportunities for achieving continual\nself-improvement of perception modules in intelligent technical systems.\n", "date": "2024-11-01", "categories": []}
{"id": "2410.23901", "title": "NeFF-BioNet: Crop Biomass Prediction from Point Cloud to Drone Imagery", "abstract": "  Crop biomass offers crucial insights into plant health and yield, making it\nessential for crop science, farming systems, and agricultural research.\nHowever, current measurement methods, which are labor-intensive, destructive,\nand imprecise, hinder large-scale quantification of this trait. To address this\nlimitation, we present a biomass prediction network (BioNet), designed for\nadaptation across different data modalities, including point clouds and drone\nimagery. Our BioNet, utilizing a sparse 3D convolutional neural network (CNN)\nand a transformer-based prediction module, processes point clouds and other 3D\ndata representations to predict biomass. To further extend BioNet for drone\nimagery, we integrate a neural feature field (NeFF) module, enabling 3D\nstructure reconstruction and the transformation of 2D semantic features from\nvision foundation models into the corresponding 3D surfaces. For the point\ncloud modality, BioNet demonstrates superior performance on two public\ndatasets, with an approximate 6.1% relative improvement (RI) over the\nstate-of-the-art. In the RGB image modality, the combination of BioNet and NeFF\nachieves a 7.9% RI. Additionally, the NeFF-based approach utilizes inexpensive,\nportable drone-mounted cameras, providing a scalable solution for large field\napplications.\n", "date": "2024-10-30", "categories": []}
{"id": "2410.19784", "title": "Enhancing Apple's Defect Classification: Insights from Visible Spectrum\n  and Narrow Spectral Band Imaging", "abstract": "  This study addresses the classification of defects in apples as a crucial\nmeasure to mitigate economic losses and optimize the food supply chain. An\ninnovative approach is employed that integrates images from the visible\nspectrum and 660 nm spectral wavelength to enhance accuracy and efficiency in\ndefect classification. The methodology is based on the use of Single-Input and\nMulti-Inputs convolutional neural networks (CNNs) to validate the proposed\nstrategies. Steps include image acquisition and preprocessing, classification\nmodel training, and performance evaluation. Results demonstrate that defect\nclassification using the 660 nm spectral wavelength reveals details not visible\nin the entire visible spectrum. It is seen that the use of the appropriate\nspectral range in the classification process is slightly superior to the entire\nvisible spectrum. The MobileNetV1 model achieves an accuracy of 98.80\\% on the\nvalidation dataset versus the 98.26\\% achieved using the entire visible\nspectrum. Conclusions highlight the potential to enhance the method by\ncapturing images with specific spectral ranges using filters, enabling more\neffective network training for classification task. These improvements could\nfurther enhance the system's capability to identify and classify defects in\napples.\n", "date": "2024-10-14", "categories": []}
{"id": "2209.14003", "title": "Vision based Crop Row Navigation under Varying Field Conditions in\n  Arable Fields", "abstract": "  Accurate crop row detection is often challenged by the varying field\nconditions present in real-world arable fields. Traditional colour based\nsegmentation is unable to cater for all such variations. The lack of\ncomprehensive datasets in agricultural environments limits the researchers from\ndeveloping robust segmentation models to detect crop rows. We present a dataset\nfor crop row detection with 11 field variations from Sugar Beet and Maize\ncrops. We also present a novel crop row detection algorithm for visual servoing\nin crop row fields. Our algorithm can detect crop rows against varying field\nconditions such as curved crop rows, weed presence, discontinuities, growth\nstages, tramlines, shadows and light levels. Our method only uses RGB images\nfrom a front-mounted camera on a Husky robot to predict crop rows. Our method\noutperformed the classic colour based crop row detection baseline. Dense weed\npresence within inter-row space and discontinuities in crop rows were the most\nchallenging field conditions for our crop row detection algorithm. Our method\ncan detect the end of the crop row and navigate the robot towards the headland\narea when it reaches the end of the crop row.\n", "date": "2022-09-28", "categories": []}
{"id": "2309.13393", "title": "AgriSORT: A Simple Online Real-time Tracking-by-Detection framework for\n  robotics in precision agriculture", "abstract": "  The problem of multi-object tracking (MOT) consists in detecting and tracking\nall the objects in a video sequence while keeping a unique identifier for each\nobject. It is a challenging and fundamental problem for robotics. In precision\nagriculture the challenge of achieving a satisfactory solution is amplified by\nextreme camera motion, sudden illumination changes, and strong occlusions. Most\nmodern trackers rely on the appearance of objects rather than motion for\nassociation, which can be ineffective when most targets are static objects with\nthe same appearance, as in the agricultural case. To this end, on the trail of\nSORT [5], we propose AgriSORT, a simple, online, real-time\ntracking-by-detection pipeline for precision agriculture based only on motion\ninformation that allows for accurate and fast propagation of tracks between\nframes. The main focuses of AgriSORT are efficiency, flexibility, minimal\ndependencies, and ease of deployment on robotic platforms. We test the proposed\npipeline on a novel MOT benchmark specifically tailored for the agricultural\ncontext, based on video sequences taken in a table grape vineyard, particularly\nchallenging due to strong self-similarity and density of the instances. Both\nthe code and the dataset are available for future comparisons.\n", "date": "2023-09-23", "categories": []}
{"id": "2410.18208", "title": "Automated Defect Detection and Grading of Piarom Dates Using Deep\n  Learning", "abstract": "  Grading and quality control of Piarom dates, a premium and high-value variety\ncultivated predominantly in Iran, present significant challenges due to the\ncomplexity and variability of defects, as well as the absence of specialized\nautomated systems tailored to this fruit. Traditional manual inspection methods\nare labor intensive, time consuming, and prone to human error, while existing\nAI-based sorting solutions are insufficient for addressing the nuanced\ncharacteristics of Piarom dates. In this study, we propose an innovative deep\nlearning framework designed specifically for the real-time detection,\nclassification, and grading of Piarom dates. Leveraging a custom dataset\ncomprising over 9,900 high-resolution images annotated across 11 distinct\ndefect categories, our framework integrates state-of-the-art object detection\nalgorithms and Convolutional Neural Networks (CNNs) to achieve high precision\nin defect identification. Furthermore, we employ advanced segmentation\ntechniques to estimate the area and weight of each date, thereby optimizing the\ngrading process according to industry standards. Experimental results\ndemonstrate that our system significantly outperforms existing methods in terms\nof accuracy and computational efficiency, making it highly suitable for\nindustrial applications requiring real-time processing. This work not only\nprovides a robust and scalable solution for automating quality control in the\nPiarom date industry but also contributes to the broader field of AI-driven\nfood inspection technologies, with potential applications across various\nagricultural products.\n", "date": "2024-10-23", "categories": []}
{"id": "2410.11108", "title": "Classifying Healthy and Defective Fruits with a Multi-Input Architecture\n  and CNN Models", "abstract": "  This study presents an investigation into the utilization of a Multi-Input\narchitecture for the classification of fruits (apples and mangoes) into healthy\nand defective states, employing both RGB and silhouette images. The primary aim\nis to enhance the accuracy of CNN models. The methodology encompasses image\nacquisition, preprocessing of datasets, training, and evaluation of two CNN\nmodels: MobileNetV2 and VGG16. Results reveal that the inclusion of silhouette\nimages alongside the Multi-Input architecture yields models with superior\nperformance compared to using only RGB images for fruit classification, whether\nhealthy or defective. Specifically, optimal results were achieved using the\nMobileNetV2 model, achieving 100\\% accuracy. This finding suggests the efficacy\nof this combined methodology in improving the precise classification of healthy\nor defective fruits, which could have significant implications for applications\nrelated to external quality inspection of fruits.\n", "date": "2024-10-14", "categories": []}
{"id": "2310.15138", "title": "Fusion-Driven Tree Reconstruction and Fruit Localization: Advancing\n  Precision in Agriculture", "abstract": "  Fruit distribution is pivotal in shaping the future of both agriculture and\nagricultural robotics, paving the way for a streamlined supply chain. This\nstudy introduces an innovative methodology that harnesses the synergy of RGB\nimagery, LiDAR, and IMU data, to achieve intricate tree reconstructions and the\npinpoint localization of fruits. Such integration not only offers insights into\nthe fruit distribution, which enhances the precision of guidance for\nagricultural robotics and automation systems, but also sets the stage for\nsimulating synthetic fruit patterns across varied tree architectures. To\nvalidate this approach, experiments have been carried out in both a controlled\nenvironment and an actual peach orchard. The results underscore the robustness\nand efficacy of this fusion-driven methodology, highlighting its potential as a\ntransformative tool for future agricultural robotics and precision farming.\n", "date": "2023-10-23", "categories": []}
{"id": "2309.05645", "title": "CitDet: A Benchmark Dataset for Citrus Fruit Detection", "abstract": "  In this letter, we present a new dataset to advance the state of the art in\ndetecting citrus fruit and accurately estimate yield on trees affected by the\nHuanglongbing (HLB) disease in orchard environments via imaging. Despite the\nfact that significant progress has been made in solving the fruit detection\nproblem, the lack of publicly available datasets has complicated direct\ncomparison of results. For instance, citrus detection has long been of interest\nto the agricultural research community, yet there is an absence of work,\nparticularly involving public datasets of citrus affected by HLB. To address\nthis issue, we enhance state-of-the-art object detection methods for use in\ntypical orchard settings. Concretely, we provide high-resolution images of\ncitrus trees located in an area known to be highly affected by HLB, along with\nhigh-quality bounding box annotations of citrus fruit. Fruit on both the trees\nand the ground are labeled to allow for identification of fruit location, which\ncontributes to advancements in yield estimation and potential measure of HLB\nimpact via fruit drop. The dataset consists of over 32,000 bounding box\nannotations for fruit instances contained in 579 high-resolution images. In\nsummary, our contributions are the following: (i) we introduce a novel dataset\nalong with baseline performance benchmarks on multiple contemporary object\ndetection algorithms, (ii) we show the ability to accurately capture fruit\nlocation on tree or on ground, and finally (ii) we present a correlation of our\nresults with yield estimations.\n", "date": "2023-09-11", "categories": []}
{"id": "2410.07528", "title": "CountMamba: Exploring Multi-directional Selective State-Space Models for\n  Plant Counting", "abstract": "  Plant counting is essential in every stage of agriculture, including seed\nbreeding, germination, cultivation, fertilization, pollination yield\nestimation, and harvesting. Inspired by the fact that humans count objects in\nhigh-resolution images by sequential scanning, we explore the potential of\nhandling plant counting tasks via state space models (SSMs) for generating\ncounting results. In this paper, we propose a new counting approach named\nCountMamba that constructs multiple counting experts to scan from various\ndirections simultaneously. Specifically, we design a Multi-directional\nState-Space Group to process the image patch sequences in multiple orders and\naim to simulate different counting experts. We also design Global-Local\nAdaptive Fusion to adaptively aggregate global features extracted from multiple\ndirections and local features extracted from the CNN branch in a sample-wise\nmanner. Extensive experiments demonstrate that the proposed CountMamba performs\ncompetitively on various plant counting tasks, including maize tassels, wheat\nears, and sorghum head counting.\n", "date": "2024-10-10", "categories": []}
{"id": "2410.04983", "title": "RoWeeder: Unsupervised Weed Mapping through Crop-Row Detection", "abstract": "  Precision agriculture relies heavily on effective weed management to ensure\nrobust crop yields. This study presents RoWeeder, an innovative framework for\nunsupervised weed mapping that combines crop-row detection with a\nnoise-resilient deep learning model. By leveraging crop-row information to\ncreate a pseudo-ground truth, our method trains a lightweight deep learning\nmodel capable of distinguishing between crops and weeds, even in the presence\nof noisy data. Evaluated on the WeedMap dataset, RoWeeder achieves an F1 score\nof 75.3, outperforming several baselines. Comprehensive ablation studies\nfurther validated the model's performance. By integrating RoWeeder with drone\ntechnology, farmers can conduct real-time aerial surveys, enabling precise weed\nmanagement across large fields. The code is available at:\n\\url{https://github.com/pasqualedem/RoWeeder}.\n", "date": "2024-10-07", "categories": []}
{"id": "2410.00503", "title": "Drone Stereo Vision for Radiata Pine Branch Detection and Distance\n  Measurement: Utilizing Deep Learning and YOLO Integration", "abstract": "  This research focuses on the development of a drone equipped with pruning\ntools and a stereo vision camera to accurately detect and measure the spatial\npositions of tree branches. YOLO is employed for branch segmentation, while two\ndepth estimation approaches, monocular and stereo, are investigated. In\ncomparison to SGBM, deep learning techniques produce more refined and accurate\ndepth maps. In the absence of ground-truth data, a fine-tuning process using\ndeep neural networks is applied to approximate optimal depth values. This\nmethodology facilitates precise branch detection and distance measurement,\naddressing critical challenges in the automation of pruning operations. The\nresults demonstrate notable advancements in both accuracy and efficiency,\nunderscoring the potential of deep learning to drive innovation and enhance\nautomation in the agricultural sector.\n", "date": "2024-10-01", "categories": []}
{"id": "2310.05762", "title": "MonoVisual3DFilter: 3D tomatoes' localisation with monocular cameras\n  using histogram filters", "abstract": "  Performing tasks in agriculture, such as fruit monitoring or harvesting,\nrequires perceiving the objects' spatial position. RGB-D cameras are limited\nunder open-field environments due to lightning interferences. So, in this\nstudy, we state to answer the research question: \"How can we use and control\nmonocular sensors to perceive objects' position in the 3D task space?\" Towards\nthis aim, we approached histogram filters (Bayesian discrete filters) to\nestimate the position of tomatoes in the tomato plant through the algorithm\nMonoVisual3DFilter. Two kernel filters were studied: the square kernel and the\nGaussian kernel. The implemented algorithm was essayed in simulation, with and\nwithout Gaussian noise and random noise, and in a testbed at laboratory\nconditions. The algorithm reported a mean absolute error lower than 10 mm in\nsimulation and 20 mm in the testbed at laboratory conditions with an assessing\ndistance of about 0.5 m. So, the results are viable for real environments and\nshould be improved at closer distances.\n", "date": "2023-10-09", "categories": []}
{"id": "2409.09673", "title": "SITSMamba for Crop Classification based on Satellite Image Time Series", "abstract": "  Satellite image time series (SITS) data provides continuous observations over\ntime, allowing for the tracking of vegetation changes and growth patterns\nthroughout the seasons and years. Numerous deep learning (DL) approaches using\nSITS for crop classification have emerged recently, with the latest approaches\nadopting Transformer for SITS classification. However, the quadratic complexity\nof self-attention in Transformer poses challenges for classifying long time\nseries. While the cutting-edge Mamba architecture has demonstrated strength in\nvarious domains, including remote sensing image interpretation, its capacity to\nlearn temporal representations in SITS data remains unexplored. Moreover, the\nexisting SITS classification methods often depend solely on crop labels as\nsupervision signals, which fails to fully exploit the temporal information. In\nthis paper, we proposed a Satellite Image Time Series Mamba (SITSMamba) method\nfor crop classification based on remote sensing time series data. The proposed\nSITSMamba contains a spatial encoder based on Convolutional Neural Networks\n(CNN) and a Mamba-based temporal encoder. To exploit richer temporal\ninformation from SITS, we design two branches of decoder used for different\ntasks. The first branch is a crop Classification Branch (CBranch), which\nincludes a ConvBlock to decode the feature to a crop map. The second branch is\na SITS Reconstruction Branch that uses a Linear layer to transform the encoded\nfeature to predict the original input values. Furthermore, we design a\nPositional Weight (PW) applied to the RBranch to help the model learn rich\nlatent knowledge from SITS. We also design two weighting factors to control the\nbalance of the two branches during training. The code of SITSMamba is available\nat: https://github.com/XiaoleiQinn/SITSMamba.\n", "date": "2024-09-15", "categories": []}
{"id": "2409.19963", "title": "A Self-attention Residual Convolutional Neural Network for Health\n  Condition Classification of Cow Teat Images", "abstract": "  Milk is a highly important consumer for Americans and the health of the cows'\nteats directly affects the quality of the milk. Traditionally, veterinarians\nmanually assessed teat health by visually inspecting teat-end hyperkeratosis\nduring the milking process which is limited in time, usually only tens of\nseconds, and weakens the accuracy of the health assessment of cows' teats.\nConvolutional neural networks (CNNs) have been used for cows' teat-end health\nassessment. However, there are challenges in using CNNs for cows' teat-end\nhealth assessment, such as complex environments, changing positions and\npostures of cows' teats, and difficulty in identifying cows' teats from images.\nTo address these challenges, this paper proposes a cows' teats self-attention\nresidual convolutional neural network (CTSAR-CNN) model that combines residual\nconnectivity and self-attention mechanisms to assist commercial farms in the\nhealth assessment of cows' teats by classifying the magnitude of teat-end\nhyperkeratosis using digital images. The results showed that upon integrating\nresidual connectivity and self-attention mechanisms, the accuracy of CTSAR-CNN\nhas been improved. This research illustrates that CTSAR-CNN can be more\nadaptable and speedy to assist veterinarians in assessing the health of cows'\nteats and ultimately benefit the dairy industry.\n", "date": "2024-09-30", "categories": []}
{"id": "2408.14206", "title": "Lemon and Orange Disease Classification using CNN-Extracted Features and\n  Machine Learning Classifier", "abstract": "  Lemons and oranges, both are the most economically significant citrus fruits\nglobally. The production of lemons and oranges is severely affected due to\ndiseases in its growth stages. Fruit quality has degraded due to the presence\nof flaws. Thus, it is necessary to diagnose the disease accurately so that we\ncan avoid major loss of lemons and oranges. To improve citrus farming, we\nproposed a disease classification approach for lemons and oranges. This\napproach would enable early disease detection and intervention, reduce yield\nlosses, and optimize resource allocation. For the initial modeling of disease\nclassification, the research uses innovative deep learning architectures such\nas VGG16, VGG19 and ResNet50. In addition, for achieving better accuracy, the\nbasic machine learning algorithms used for classification problems include\nRandom Forest, Naive Bayes, K-Nearest Neighbors (KNN) and Logistic Regression.\nThe lemon and orange fruits diseases are classified more accurately (95.0% for\nlemon and 99.69% for orange) by the model. The model's base features were\nextracted from the ResNet50 pre-trained model and the diseases are classified\nby the Logistic Regression which beats the performance given by VGG16 and VGG19\nfor other classifiers. Experimental outcomes show that the proposed model also\noutperforms existing models in which most of them classified the diseases using\nthe Softmax classifier without using any individual classifiers.\n", "date": "2024-08-26", "categories": []}
{"id": "2408.06190", "title": "FruitNeRF: A Unified Neural Radiance Field based Fruit Counting\n  Framework", "abstract": "  We introduce FruitNeRF, a unified novel fruit counting framework that\nleverages state-of-the-art view synthesis methods to count any fruit type\ndirectly in 3D. Our framework takes an unordered set of posed images captured\nby a monocular camera and segments fruit in each image. To make our system\nindependent of the fruit type, we employ a foundation model that generates\nbinary segmentation masks for any fruit. Utilizing both modalities, RGB and\nsemantic, we train a semantic neural radiance field. Through uniform volume\nsampling of the implicit Fruit Field, we obtain fruit-only point clouds. By\napplying cascaded clustering on the extracted point cloud, our approach\nachieves precise fruit count.The use of neural radiance fields provides\nsignificant advantages over conventional methods such as object tracking or\noptical flow, as the counting itself is lifted into 3D. Our method prevents\ndouble counting fruit and avoids counting irrelevant fruit.We evaluate our\nmethodology using both real-world and synthetic datasets. The real-world\ndataset consists of three apple trees with manually counted ground truths, a\nbenchmark apple dataset with one row and ground truth fruit location, while the\nsynthetic dataset comprises various fruit types including apple, plum, lemon,\npear, peach, and mango.Additionally, we assess the performance of fruit\ncounting using the foundation model compared to a U-Net.\n", "date": "2024-08-12", "categories": []}
{"id": "2409.17526", "title": "Drone Stereo Vision for Radiata Pine Branch Detection and Distance\n  Measurement: Integrating SGBM and Segmentation Models", "abstract": "  Manual pruning of radiata pine trees presents significant safety risks due to\ntheir substantial height and the challenging terrains in which they thrive. To\naddress these risks, this research proposes the development of a drone-based\npruning system equipped with specialized pruning tools and a stereo vision\ncamera, enabling precise detection and trimming of branches. Deep learning\nalgorithms, including YOLO and Mask R-CNN, are employed to ensure accurate\nbranch detection, while the Semi-Global Matching algorithm is integrated to\nprovide reliable distance estimation. The synergy between these techniques\nfacilitates the precise identification of branch locations and enables\nefficient, targeted pruning. Experimental results demonstrate that the combined\nimplementation of YOLO and SGBM enables the drone to accurately detect branches\nand measure their distances from the drone. This research not only improves the\nsafety and efficiency of pruning operations but also makes a significant\ncontribution to the advancement of drone technology in the automation of\nagricultural and forestry practices, laying a foundational framework for\nfurther innovations in environmental management.\n", "date": "2024-09-26", "categories": []}
{"id": "2409.16412", "title": "Vision-based Xylem Wetness Classification in Stem Water Potential\n  Determination", "abstract": "  Water is often overused in irrigation, making efficient management of it\ncrucial. Precision Agriculture emphasizes tools like stem water potential (SWP)\nanalysis for better plant status determination. However, such tools often\nrequire labor-intensive in-situ sampling. Automation and machine learning can\nstreamline this process and enhance outcomes. This work focused on automating\nstem detection and xylem wetness classification using the Scholander Pressure\nChamber, a widely used but demanding method for SWP measurement. The aim was to\nrefine stem detection and develop computer-vision-based methods to better\nclassify water emergence at the xylem. To this end, we collected and manually\nannotated video data, applying vision- and learning-based methods for detection\nand classification. Additionally, we explored data augmentation and fine-tuned\nparameters to identify the most effective models. The identified\nbest-performing models for stem detection and xylem wetness classification were\nevaluated end-to-end over 20 SWP measurements. Learning-based stem detection\nvia YOLOv8n combined with ResNet50-based classification achieved a Top-1\naccuracy of 80.98%, making it the best-performing approach for xylem wetness\nclassification.\n", "date": "2024-09-24", "categories": []}
{"id": "2409.17119", "title": "Small data deep learning methodology for in-field disease detection", "abstract": "  Early detection of diseases in crops is essential to prevent harvest losses\nand improve the quality of the final product. In this context, the combination\nof machine learning and proximity sensors is emerging as a technique capable of\nachieving this detection efficiently and effectively. For example, this machine\nlearning approach has been applied to potato crops -- to detect late blight\n(Phytophthora infestans) -- and grapevine crops -- to detect downy mildew.\nHowever, most of these AI models found in the specialised literature have been\ndeveloped using leaf-by-leaf images taken in the lab, which does not represent\nfield conditions and limits their applicability.\n  In this study, we present the first machine learning model capable of\ndetecting mild symptoms of late blight in potato crops through the analysis of\nhigh-resolution RGB images captured directly in the field, overcoming the\nlimitations of other publications in the literature and presenting real-world\napplicability. Our proposal exploits the availability of high-resolution images\nvia the concept of patching, and is based on deep convolutional neural networks\nwith a focal loss function, which makes the model to focus on the complex\npatterns that arise in field conditions. Additionally, we present a data\naugmentation scheme that facilitates the training of these neural networks with\nfew high-resolution images, which allows for development of models under the\nsmall data paradigm.\n  Our model correctly detects all cases of late blight in the test dataset,\ndemonstrating a high level of accuracy and effectiveness in identifying early\nsymptoms. These promising results reinforce the potential use of machine\nlearning for the early detection of diseases and pests in agriculture, enabling\nbetter treatment and reducing their impact on crops.\n", "date": "2024-09-25", "categories": []}
{"id": "2409.15679", "title": "PDT: Uav Target Detection Dataset for Pests and Diseases Tree", "abstract": "  UAVs emerge as the optimal carriers for visual weed iden?tification and\nintegrated pest and disease management in crops. How?ever, the absence of\nspecialized datasets impedes the advancement of model development in this\ndomain. To address this, we have developed the Pests and Diseases Tree dataset\n(PDT dataset). PDT dataset repre?sents the first high-precision UAV-based\ndataset for targeted detection of tree pests and diseases, which is collected\nin real-world operational environments and aims to fill the gap in available\ndatasets for this field. Moreover, by aggregating public datasets and network\ndata, we further introduced the Common Weed and Crop dataset (CWC dataset) to\nad?dress the challenge of inadequate classification capabilities of test models\nwithin datasets for this field. Finally, we propose the YOLO-Dense Pest\n(YOLO-DP) model for high-precision object detection of weed, pest, and disease\ncrop images. We re-evaluate the state-of-the-art detection models with our\nproposed PDT dataset and CWC dataset, showing the completeness of the dataset\nand the effectiveness of the YOLO-DP. The proposed PDT dataset, CWC dataset,\nand YOLO-DP model are pre?sented at\nhttps://github.com/RuiXing123/PDT_CWC_YOLO-DP.\n", "date": "2024-09-24", "categories": []}
{"id": "2409.16213", "title": "Deep Learning for Precision Agriculture: Post-Spraying Evaluation and\n  Deposition Estimation", "abstract": "  Precision spraying evaluation requires automation primarily in post-spraying\nimagery. In this paper we propose an eXplainable Artificial Intelligence (XAI)\ncomputer vision pipeline to evaluate a precision spraying system post-spraying\nwithout the need for traditional agricultural methods. The developed system can\nsemantically segment potential targets such as lettuce, chickweed, and\nmeadowgrass and correctly identify if targets have been sprayed. Furthermore,\nthis pipeline evaluates using a domain-specific Weakly Supervised Deposition\nEstimation task, allowing for class-specific quantification of spray deposit\nweights in {\\mu}L. Estimation of coverage rates of spray deposition in a\nclass-wise manner allows for further understanding of effectiveness of\nprecision spraying systems. Our study evaluates different Class Activation\nMapping techniques, namely AblationCAM and ScoreCAM, to determine which is more\neffective and interpretable for these tasks. In the pipeline, inference-only\nfeature fusion is used to allow for further interpretability and to enable the\nautomation of precision spraying evaluation post-spray. Our findings indicate\nthat a Fully Convolutional Network with an EfficientNet-B0 backbone and\ninference-only feature fusion achieves an average absolute difference in\ndeposition values of 156.8 {\\mu}L across three classes in our test set. The\ndataset curated in this paper is publicly available at\nhttps://github.com/Harry-Rogers/PSIE\n", "date": "2024-09-24", "categories": []}
{"id": "2409.13330", "title": "Enhancing Fruit and Vegetable Detection in Unconstrained Environment\n  with a Novel Dataset", "abstract": "  Automating the detection of fruits and vegetables using computer vision is\nessential for modernizing agriculture, improving efficiency, ensuring food\nquality, and contributing to technologically advanced and sustainable farming\npractices. This paper presents an end-to-end pipeline for detecting and\nlocalizing fruits and vegetables in real-world scenarios. To achieve this, we\nhave curated a dataset named FRUVEG67 that includes images of 67 classes of\nfruits and vegetables captured in unconstrained scenarios, with only a few\nmanually annotated samples per class. We have developed a semi-supervised data\nannotation algorithm (SSDA) that generates bounding boxes for objects to label\nthe remaining non-annotated images. For detection, we introduce the Fruit and\nVegetable Detection Network (FVDNet), an ensemble version of YOLOv7 featuring\nthree distinct grid configurations. We employ an averaging approach for\nbounding-box prediction and a voting mechanism for class prediction. We have\nintegrated Jensen-Shannon divergence (JSD) in conjunction with focal loss to\nbetter detect smaller objects. Our experimental results highlight the\nsuperiority of FVDNet compared to previous versions of YOLO, showcasing\nremarkable improvements in detection and localization performance. We achieved\nan impressive mean average precision (mAP) score of 0.78 across all classes.\nFurthermore, we evaluated the efficacy of FVDNet using open-category\nrefrigerator images, where it demonstrates promising results.\n", "date": "2024-09-20", "categories": []}
{"id": "2409.12350", "title": "Advancing Cucumber Disease Detection in Agriculture through Machine\n  Vision and Drone Technology", "abstract": "  This study uses machine vision and drone technologies to propose a unique\nmethod for the diagnosis of cucumber disease in agriculture. The backbone of\nthis research is a painstakingly curated dataset of hyperspectral photographs\nacquired under genuine field conditions. Unlike earlier datasets, this study\nincluded a wide variety of illness types, allowing for precise early-stage\ndetection. The model achieves an excellent 87.5\\% accuracy in distinguishing\neight unique cucumber illnesses after considerable data augmentation. The\nincorporation of drone technology for high-resolution images improves disease\nevaluation. This development has enormous potential for improving crop\nmanagement, lowering labor costs, and increasing agricultural productivity.\nThis research, which automates disease detection, represents a significant step\ntoward a more efficient and sustainable agricultural future.\n", "date": "2024-09-18", "categories": []}
{"id": "2409.10445", "title": "Deep-Wide Learning Assistance for Insect Pest Classification", "abstract": "  Accurate insect pest recognition plays a critical role in agriculture. It is\na challenging problem due to the intricate characteristics of insects. In this\npaper, we present DeWi, novel learning assistance for insect pest\nclassification. With a one-stage and alternating training strategy, DeWi\nsimultaneously improves several Convolutional Neural Networks in two\nperspectives: discrimination (by optimizing a triplet margin loss in a\nsupervised training manner) and generalization (via data augmentation). From\nthat, DeWi can learn discriminative and in-depth features of insect pests\n(deep) yet still generalize well to a large number of insect categories (wide).\nExperimental results show that DeWi achieves the highest performances on two\ninsect pest classification benchmarks (76.44\\% accuracy on the IP102 dataset\nand 99.79\\% accuracy on the D0 dataset, respectively). In addition, extensive\nevaluations and ablation studies are conducted to thoroughly investigate our\nDeWi and demonstrate its superiority. Our source code is available at\nhttps://github.com/toannguyen1904/DeWi.\n", "date": "2024-09-16", "categories": []}
{"id": "2404.11256", "title": "MMCBE: Multi-modality Dataset for Crop Biomass Prediction and Beyond", "abstract": "  Crop biomass, a critical indicator of plant growth, health, and productivity,\nis invaluable for crop breeding programs and agronomic research. However, the\naccurate and scalable quantification of crop biomass remains inaccessible due\nto limitations in existing measurement methods. One of the obstacles impeding\nthe advancement of current crop biomass prediction methodologies is the\nscarcity of publicly available datasets. Addressing this gap, we introduce a\nnew dataset in this domain, i.e. Multi-modality dataset for crop biomass\nestimation (MMCBE). Comprising 216 sets of multi-view drone images, coupled\nwith LiDAR point clouds, and hand-labelled ground truth, MMCBE represents the\nfirst multi-modality one in the field. This dataset aims to establish benchmark\nmethods for crop biomass quantification and foster the development of\nvision-based approaches. We have rigorously evaluated state-of-the-art crop\nbiomass estimation methods using MMCBE and ventured into additional potential\napplications, such as 3D crop reconstruction from drone imagery and novel-view\nrendering. With this publication, we are making our comprehensive dataset\navailable to the broader community.\n", "date": "2024-04-17", "categories": []}
{"id": "2409.04060", "title": "D4: Text-guided diffusion model-based domain adaptive data augmentation\n  for vineyard shoot detection", "abstract": "  In an agricultural field, plant phenotyping using object detection models is\ngaining attention. However, collecting the training data necessary to create\ngeneric and high-precision models is extremely challenging due to the\ndifficulty of annotation and the diversity of domains. Furthermore, it is\ndifficult to transfer training data across different crops, and although\nmachine learning models effective for specific environments, conditions, or\ncrops have been developed, they cannot be widely applied in actual fields. In\nthis study, we propose a generative data augmentation method (D4) for vineyard\nshoot detection. D4 uses a pre-trained text-guided diffusion model based on a\nlarge number of original images culled from video data collected by unmanned\nground vehicles or other means, and a small number of annotated datasets. The\nproposed method generates new annotated images with background information\nadapted to the target domain while retaining annotation information necessary\nfor object detection. In addition, D4 overcomes the lack of training data in\nagriculture, including the difficulty of annotation and diversity of domains.\nWe confirmed that this generative data augmentation method improved the mean\naverage precision by up to 28.65% for the BBox detection task and the average\nprecision by up to 13.73% for the keypoint detection task for vineyard shoot\ndetection. Our generative data augmentation method D4 is expected to\nsimultaneously solve the cost and domain diversity issues of training data\ngeneration in agriculture and improve the generalization performance of\ndetection models.\n", "date": "2024-09-06", "categories": []}
{"id": "2409.04068", "title": "Site-Specific Color Features of Green Coffee Beans", "abstract": "  Coffee is one of the most valuable primary commodities. Despite this, the\ncommon selection technique of green coffee beans relies on personnel visual\ninspection, which is labor-intensive and subjective. Therefore, an efficient\nway to evaluate the quality of beans is needed. In this paper, we demonstrate a\nsite-independent approach to find site-specific color features of the seed coat\nin qualified green coffee beans. We then propose two evaluation schemes for\ngreen coffee beans based on this site-specific color feature of qualified\nbeans. Due to the site-specific properties of these color features, machine\nlearning classifiers indicate that compared with the existing evaluation\nschemes of beans, our evaluation schemes have the advantages of being simple,\nhaving less computational costs, and having universal applicability. Finally,\nthis site-specific color feature can distinguish qualified beans from different\ngrowing sites. Moreover, this function can prevent cheating in the coffee\nbusiness and is unique to our evaluation scheme of beans.\n", "date": "2024-09-06", "categories": []}
{"id": "2409.00395", "title": "Self-supervised Fusarium Head Blight Detection with Hyperspectral Image\n  and Feature Mining", "abstract": "  Fusarium Head Blight (FHB) is a serious fungal disease affecting wheat\n(including durum), barley, oats, other small cereal grains, and corn. Effective\nmonitoring and accurate detection of FHB are crucial to ensuring stable and\nreliable food security. Traditionally, trained agronomists and surveyors\nperform manual identification, a method that is labor-intensive, impractical,\nand challenging to scale. With the advancement of deep learning and\nHyper-spectral Imaging (HSI) and Remote Sensing (RS) technologies, employing\ndeep learning, particularly Convolutional Neural Networks (CNNs), has emerged\nas a promising solution. Notably, wheat infected with serious FHB may exhibit\nsignificant differences on the spectral compared to mild FHB one, which is\nparticularly advantageous for hyperspectral image-based methods. In this study,\nwe propose a self-unsupervised classification method based on HSI endmember\nextraction strategy and top-K bands selection, designed to analyze material\nsignatures in HSIs to derive discriminative feature representations. This\napproach does not require expensive device or complicate algorithm design,\nmaking it more suitable for practical uses. Our method has been effectively\nvalidated in the Beyond Visible Spectrum: AI for Agriculture Challenge 2024.\nThe source code is easy to reproduce and available at\n{https://github.com/VanLinLin/Automated-Crop-Disease-Diagnosis-from-Hyperspectral-Imagery-3rd}.\n", "date": "2024-08-31", "categories": []}
{"id": "2409.00513", "title": "Plant detection from ultra high resolution remote sensing images: A\n  Semantic Segmentation approach based on fuzzy loss", "abstract": "  In this study, we tackle the challenge of identifying plant species from\nultra high resolution (UHR) remote sensing images. Our approach involves\nintroducing an RGB remote sensing dataset, characterized by millimeter-level\nspatial resolution, meticulously curated through several field expeditions\nacross a mountainous region in France covering various landscapes. The task of\nplant species identification is framed as a semantic segmentation problem for\nits practical and efficient implementation across vast geographical areas.\nHowever, when dealing with segmentation masks, we confront instances where\ndistinguishing boundaries between plant species and their background is\nchallenging. We tackle this issue by introducing a fuzzy loss within the\nsegmentation model. Instead of utilizing one-hot encoded ground truth (GT), our\nmodel incorporates Gaussian filter refined GT, introducing stochasticity during\ntraining. First experimental results obtained on both our UHR dataset and a\npublic dataset are presented, showing the relevance of the proposed\nmethodology, as well as the need for future improvement.\n", "date": "2024-08-31", "categories": []}
{"id": "2304.07853", "title": "AI driven shadow model detection in agropv farms", "abstract": "  Agro-photovoltaic (APV) is a growing farming practice that combines\nagriculture and solar photovoltaic projects within the same area. This emerging\nmarket is expected to experience significant growth in the next few years, with\na projected investment of $9 billion in 2030. Identifying shadows is crucial to\nunderstanding the APV environment, as they impact plant growth, microclimate,\nand evapotranspiration. In this study, we use state-of-the-art CNN and\nGAN-based neural networks to detect shadows in agro-PV farms, demonstrating\ntheir effectiveness. However, challenges remain, including partial shadowing\nfrom moving objects and real-time monitoring. Future research should focus on\ndeveloping more sophisticated neural network-based shadow detection algorithms\nand integrating them with control systems for APV farms. Overall, shadow\ndetection is crucial to increase productivity and profitability while\nsupporting the environment, soil, and farmers.\n", "date": "2023-04-16", "categories": []}
{"id": "2405.01310", "title": "Overcoming LLM Challenges using RAG-Driven Precision in Coffee Leaf\n  Disease Remediation", "abstract": "  This research introduces an innovative AI-driven precision agriculture\nsystem, leveraging YOLOv8 for disease identification and Retrieval Augmented\nGeneration (RAG) for context-aware diagnosis. Focused on addressing the\nchallenges of diseases affecting the coffee production sector in Karnataka, The\nsystem integrates sophisticated object detection techniques with language\nmodels to address the inherent constraints associated with Large Language\nModels (LLMs). Our methodology not only tackles the issue of hallucinations in\nLLMs, but also introduces dynamic disease identification and remediation\nstrategies. Real-time monitoring, collaborative dataset expansion, and\norganizational involvement ensure the system's adaptability in diverse\nagricultural settings. The effect of the suggested system extends beyond\nautomation, aiming to secure food supplies, protect livelihoods, and promote\neco-friendly farming practices. By facilitating precise disease identification,\nthe system contributes to sustainable and environmentally conscious\nagriculture, reducing reliance on pesticides. Looking to the future, the\nproject envisions continuous development in RAG-integrated object detection\nsystems, emphasizing scalability, reliability, and usability. This research\nstrives to be a beacon for positive change in agriculture, aligning with global\nefforts toward sustainable and technologically enhanced food production.\n", "date": "2024-05-02", "categories": []}
{"id": "2408.16472", "title": "Creating a Segmented Pointcloud of Grapevines by Combining Multiple\n  Viewpoints Through Visual Odometry", "abstract": "  Grapevine winter pruning is a labor-intensive and repetitive process that\nsignificantly influences the quality and quantity of the grape harvest and\nproduced wine of the following season. It requires a careful and expert\ndetection of the point to be cut. Because of its complexity, repetitive nature\nand time constraint, the task requires skilled labor that needs to be trained.\nThis extended abstract presents the computer vision pipeline employed in\nproject Vinum, using detectron2 as a segmentation network and keypoint visual\nodometry to merge different observation into a single pointcloud used to make\ninformed pruning decisions.\n", "date": "2024-08-29", "categories": []}
{"id": "2408.13160", "title": "KonvLiNA: Integrating Kolmogorov-Arnold Network with Linear Nystr\\\"om\n  Attention for feature fusion in Crop Field Detection", "abstract": "  Crop field detection is a critical component of precision agriculture,\nessential for optimizing resource allocation and enhancing agricultural\nproductivity. This study introduces KonvLiNA, a novel framework that integrates\nConvolutional Kolmogorov-Arnold Networks (cKAN) with Nystr\\\"om attention\nmechanisms for effective crop field detection. Leveraging KAN adaptive\nactivation functions and the efficiency of Nystr\\\"om attention in handling\nlargescale data, KonvLiNA significantly enhances feature extraction, enabling\nthe model to capture intricate patterns in complex agricultural environments.\nExperimental results on rice crop dataset demonstrate KonvLiNA superiority over\nstate-of-the-art methods, achieving a 0.415 AP and 0.459 AR with the Swin-L\nbackbone, outperforming traditional YOLOv8 by significant margins.\nAdditionally, evaluation on the COCO dataset showcases competitive performance\nacross small, medium, and large objects, highlighting KonvLiNA efficacy in\ndiverse agricultural settings. This work highlights the potential of hybrid KAN\nand attention mechanisms for advancing precision agriculture through improved\ncrop field detection and management.\n", "date": "2024-08-23", "categories": []}
{"id": "2408.12426", "title": "Enhanced Infield Agriculture with Interpretable Machine Learning\n  Approaches for Crop Classification", "abstract": "  The increasing popularity of Artificial Intelligence in recent years has led\nto a surge in interest in image classification, especially in the agricultural\nsector. With the help of Computer Vision, Machine Learning, and Deep Learning,\nthe sector has undergone a significant transformation, leading to the\ndevelopment of new techniques for crop classification in the field. Despite the\nextensive research on various image classification techniques, most have\nlimitations such as low accuracy, limited use of data, and a lack of reporting\nmodel size and prediction. The most significant limitation of all is the need\nfor model explainability. This research evaluates four different approaches for\ncrop classification, namely traditional ML with handcrafted feature extraction\nmethods like SIFT, ORB, and Color Histogram; Custom Designed CNN and\nestablished DL architecture like AlexNet; transfer learning on five models\npre-trained using ImageNet such as EfficientNetV2, ResNet152V2, Xception,\nInception-ResNetV2, MobileNetV3; and cutting-edge foundation models like YOLOv8\nand DINOv2, a self-supervised Vision Transformer Model. All models performed\nwell, but Xception outperformed all of them in terms of generalization,\nachieving 98% accuracy on the test data, with a model size of 80.03 MB and a\nprediction time of 0.0633 seconds. A key aspect of this research was the\napplication of Explainable AI to provide the explainability of all the models.\nThis journal presents the explainability of Xception model with LIME, SHAP, and\nGradCAM, ensuring transparency and trustworthiness in the models' predictions.\nThis study highlights the importance of selecting the right model according to\ntask-specific needs. It also underscores the important role of explainability\nin deploying AI in agriculture, providing insightful information to help\nenhance AI-driven crop management strategies.\n", "date": "2024-08-22", "categories": []}
{"id": "2408.05661", "title": "Performance Evaluation of YOLOv8 Model Configurations, for Instance\n  Segmentation of Strawberry Fruit Development Stages in an Open Field\n  Environment", "abstract": "  Accurate identification of strawberries during their maturing stages is\ncrucial for optimizing yield management, and pest control, and making informed\ndecisions related to harvest and post-harvest logistics. This study evaluates\nthe performance of YOLOv8 model configurations for instance segmentation of\nstrawberries into ripe and unripe stages in an open field environment. The\nYOLOv8n model demonstrated superior segmentation accuracy with a mean Average\nPrecision (mAP) of 80.9\\%, outperforming other YOLOv8 configurations. In terms\nof inference speed, YOLOv8n processed images at 12.9 milliseconds, while\nYOLOv8s, the least-performing model, processed at 22.2 milliseconds. Over 86\ntest images with 348 ground truth labels, YOLOv8n detected 235 ripe fruit\nclasses and 51 unripe fruit classes out of 251 ground truth ripe fruits and 97\nunripe ground truth labels, respectively. In comparison, YOLOv8s detected 204\nripe fruits and 37 unripe fruits. Overall, YOLOv8n achieved the fastest\ninference speed of 24.2 milliseconds, outperforming YOLOv8s, YOLOv8m, YOLOv8l,\nand YOLOv8x, which processed images at 33.0 milliseconds, 44.3 milliseconds,\n53.6 milliseconds, and 62.5 milliseconds, respectively. These results\nunderscore the potential of advanced object segmentation algorithms to address\ncomplex visual recognition tasks in open-field agriculture effectively to\naddress complex visual recognition tasks in open-field agriculture effectively.\n", "date": "2024-08-11", "categories": []}
{"id": "2401.07398", "title": "Cross Domain Early Crop Mapping using CropSTGAN", "abstract": "  Driven by abundant satellite imagery, machine learning-based approaches have\nrecently been promoted to generate high-resolution crop cultivation maps to\nsupport many agricultural applications. One of the major challenges faced by\nthese approaches is the limited availability of ground truth labels. In the\nabsence of ground truth, existing work usually adopts the \"direct transfer\nstrategy\" that trains a classifier using historical labels collected from other\nregions and then applies the trained model to the target region. Unfortunately,\nthe spectral features of crops exhibit inter-region and inter-annual\nvariability due to changes in soil composition, climate conditions, and crop\nprogress, the resultant models perform poorly on new and unseen regions or\nyears. Despite recent efforts, such as the application of the deep adaptation\nneural network (DANN) model structure in the deep adaptation crop\nclassification network (DACCN), to tackle the above cross-domain challenges,\ntheir effectiveness diminishes significantly when there is a large\ndissimilarity between the source and target regions. This paper introduces the\nCrop Mapping Spectral-temporal Generative Adversarial Neural Network\n(CropSTGAN), a novel solution for cross-domain challenges, that doesn't require\ntarget domain labels. CropSTGAN learns to transform the target domain's\nspectral features to those of the source domain, effectively bridging large\ndissimilarities. Additionally, it employs an identity loss to maintain the\nintrinsic local structure of the data. Comprehensive experiments across various\nregions and years demonstrate the benefits and effectiveness of the proposed\napproach. In experiments, CropSTGAN is benchmarked against various\nstate-of-the-art (SOTA) methods. Notably, CropSTGAN significantly outperforms\nthese methods in scenarios with large data distribution dissimilarities between\nthe target and source domains.\n", "date": "2024-01-15", "categories": []}
{"id": "2408.02053", "title": "PanicleNeRF: low-cost, high-precision in-field phenotypingof rice\n  panicles with smartphone", "abstract": "  The rice panicle traits significantly influence grain yield, making them a\nprimary target for rice phenotyping studies. However, most existing techniques\nare limited to controlled indoor environments and difficult to capture the rice\npanicle traits under natural growth conditions. Here, we developed PanicleNeRF,\na novel method that enables high-precision and low-cost reconstruction of rice\npanicle three-dimensional (3D) models in the field using smartphone. The\nproposed method combined the large model Segment Anything Model (SAM) and the\nsmall model You Only Look Once version 8 (YOLOv8) to achieve high-precision\nsegmentation of rice panicle images. The NeRF technique was then employed for\n3D reconstruction using the images with 2D segmentation. Finally, the resulting\npoint clouds are processed to successfully extract panicle traits. The results\nshow that PanicleNeRF effectively addressed the 2D image segmentation task,\nachieving a mean F1 Score of 86.9% and a mean Intersection over Union (IoU) of\n79.8%, with nearly double the boundary overlap (BO) performance compared to\nYOLOv8. As for point cloud quality, PanicleNeRF significantly outperformed\ntraditional SfM-MVS (structure-from-motion and multi-view stereo) methods, such\nas COLMAP and Metashape. The panicle length was then accurately extracted with\nthe rRMSE of 2.94% for indica and 1.75% for japonica rice. The panicle volume\nestimated from 3D point clouds strongly correlated with the grain number (R2 =\n0.85 for indica and 0.82 for japonica) and grain mass (0.80 for indica and 0.76\nfor japonica). This method provides a low-cost solution for high-throughput\nin-field phenotyping of rice panicles, accelerating the efficiency of rice\nbreeding.\n", "date": "2024-08-04", "categories": []}
{"id": "2408.01244", "title": "Automated Classification of Dry Bean Varieties Using XGBoost and SVM\n  Models", "abstract": "  This paper presents a comparative study on the automated classification of\nseven different varieties of dry beans using machine learning models.\nLeveraging a dataset of 12,909 dry bean samples, reduced from an initial 13,611\nthrough outlier removal and feature extraction, we applied Principal Component\nAnalysis (PCA) for dimensionality reduction and trained two multiclass\nclassifiers: XGBoost and Support Vector Machine (SVM). The models were\nevaluated using nested cross-validation to ensure robust performance assessment\nand hyperparameter tuning. The XGBoost and SVM models achieved overall correct\nclassification rates of 94.00% and 94.39%, respectively. The results underscore\nthe efficacy of these machine learning approaches in agricultural applications,\nparticularly in enhancing the uniformity and efficiency of seed classification.\nThis study contributes to the growing body of work on precision agriculture,\ndemonstrating that automated systems can significantly support seed quality\ncontrol and crop yield optimization. Future work will explore incorporating\nmore diverse datasets and advanced algorithms to further improve classification\naccuracy.\n", "date": "2024-08-02", "categories": []}
{"id": "2308.05846", "title": "Seed Kernel Counting using Domain Randomization and Object Tracking\n  Neural Networks", "abstract": "  High-throughput phenotyping (HTP) of seeds, also known as seed phenotyping,\nis the comprehensive assessment of complex seed traits such as growth,\ndevelopment, tolerance, resistance, ecology, yield, and the measurement of\nparameters that form more complex traits. One of the key aspects of seed\nphenotyping is cereal yield estimation that the seed production industry relies\nupon to conduct their business. While mechanized seed kernel counters are\navailable in the market currently, they are often priced high and sometimes\noutside the range of small scale seed production firms' affordability. The\ndevelopment of object tracking neural network models such as You Only Look Once\n(YOLO) enables computer scientists to design algorithms that can estimate\ncereal yield inexpensively. The key bottleneck with neural network models is\nthat they require a plethora of labelled training data before they can be put\nto task. We demonstrate that the use of synthetic imagery serves as a feasible\nsubstitute to train neural networks for object tracking that includes the tasks\nof object classification and detection. Furthermore, we propose a seed kernel\ncounter that uses a low-cost mechanical hopper, trained YOLOv8 neural network\nmodel, and object tracking algorithms on StrongSORT and ByteTrack to estimate\ncereal yield from videos. The experiment yields a seed kernel count with an\naccuracy of 95.2\\% and 93.2\\% for Soy and Wheat respectively using the\nStrongSORT algorithm, and an accuray of 96.8\\% and 92.4\\% for Soy and Wheat\nrespectively using the ByteTrack algorithm.\n", "date": "2023-08-10", "categories": []}
{"id": "2405.12313", "title": "Deep learning-based hyperspectral image reconstruction for quality\n  assessment of agro-product", "abstract": "  Hyperspectral imaging (HSI) has recently emerged as a promising tool for many\nagricultural applications; however, the technology cannot be directly used in a\nreal-time system due to the extensive time needed to process large volumes of\ndata. Consequently, the development of a simple, compact, and cost-effective\nimaging system is not possible with the current HSI systems. Therefore, the\noverall goal of this study was to reconstruct hyperspectral images from RGB\nimages through deep learning for agricultural applications. Specifically, this\nstudy used Hyperspectral Convolutional Neural Network - Dense (HSCNN-D) to\nreconstruct hyperspectral images from RGB images for predicting soluble solid\ncontent (SSC) in sweet potatoes. The algorithm accurately reconstructed the\nhyperspectral images from RGB images, with the resulting spectra closely\nmatching the ground-truth. The partial least squares regression (PLSR) model\nbased on reconstructed spectra outperformed the model using the full spectral\nrange, demonstrating its potential for SSC prediction in sweet potatoes. These\nfindings highlight the potential of deep learning-based hyperspectral image\nreconstruction as a low-cost, efficient tool for various agricultural uses.\n", "date": "2024-05-20", "categories": []}
{"id": "2309.00817", "title": "Soil Image Segmentation Based on Mask R-CNN", "abstract": "  The complex background in the soil image collected in the field natural\nenvironment will affect the subsequent soil image recognition based on machine\nvision. Segmenting the soil center area from the soil image can eliminate the\ninfluence of the complex background, which is an important preprocessing work\nfor subsequent soil image recognition. For the first time, the deep learning\nmethod was applied to soil image segmentation, and the Mask R-CNN model was\nselected to complete the positioning and segmentation of soil images. Construct\na soil image dataset based on the collected soil images, use the EISeg\nannotation tool to mark the soil area as soil, and save the annotation\ninformation; train the Mask R-CNN soil image instance segmentation model. The\ntrained model can obtain accurate segmentation results for soil images, and can\nshow good performance on soil images collected in different environments; the\ntrained instance segmentation model has a loss value of 0.1999 in the training\nset, and the mAP of the validation set segmentation (IoU=0.5) is 0.8804, and it\ntakes only 0.06s to complete image segmentation based on GPU acceleration,\nwhich can meet the real-time segmentation and detection of soil images in the\nfield under natural conditions. You can get our code in the Conclusions. The\nhomepage is https://github.com/YidaMyth.\n", "date": "2023-09-02", "categories": []}
{"id": "2407.17906", "title": "Hierarchical Object Detection and Recognition Framework for Practical\n  Plant Disease Diagnosis", "abstract": "  Recently, object detection methods (OD; e.g., YOLO-based models) have been\nwidely utilized in plant disease diagnosis. These methods demonstrate\nrobustness to distance variations and excel at detecting small lesions compared\nto classification methods (CL; e.g., CNN models). However, there are issues\nsuch as low diagnostic performance for hard-to-detect diseases and high\nlabeling costs. Additionally, since healthy cases cannot be explicitly trained,\nthere is a risk of false positives. We propose the Hierarchical object\ndetection and recognition framework (HODRF), a sophisticated and highly\nintegrated two-stage system that combines the strengths of both OD and CL for\nplant disease diagnosis. In the first stage, HODRF uses OD to identify regions\nof interest (ROIs) without specifying the disease. In the second stage, CL\ndiagnoses diseases surrounding the ROIs. HODRF offers several advantages: (1)\nSince OD detects only one type of ROI, HODRF can detect diseases with limited\ntraining images by leveraging its ability to identify other lesions. (2) While\nOD over-detects healthy cases, HODRF significantly reduces these errors by\nusing CL in the second stage. (3) CL's accuracy improves in HODRF as it\nidentifies diagnostic targets given as ROIs, making it less vulnerable to size\nchanges. (4) HODRF benefits from CL's lower annotation costs, allowing it to\nlearn from a larger number of images. We implemented HODRF using YOLOv7 for OD\nand EfficientNetV2 for CL and evaluated its performance on a large-scale\ndataset (4 crops, 20 diseased and healthy classes, 281K images). HODRF\noutperformed YOLOv7 alone by 5.8 to 21.5 points on healthy data and 0.6 to 7.5\npoints on macro F1 scores, and it improved macro F1 by 1.1 to 7.2 points over\nEfficientNetV2.\n", "date": "2024-07-25", "categories": []}
{"id": "2306.04557", "title": "PhenoBench -- A Large Dataset and Benchmarks for Semantic Image\n  Interpretation in the Agricultural Domain", "abstract": "  The production of food, feed, fiber, and fuel is a key task of agriculture,\nwhich has to cope with many challenges in the upcoming decades, e.g., a higher\ndemand, climate change, lack of workers, and the availability of arable land.\nVision systems can support making better and more sustainable field management\ndecisions, but also support the breeding of new crop varieties by allowing\ntemporally dense and reproducible measurements. Recently, agricultural robotics\ngot an increasing interest in the vision and robotics communities since it is a\npromising avenue for coping with the aforementioned lack of workers and\nenabling more sustainable production. While large datasets and benchmarks in\nother domains are readily available and enable significant progress,\nagricultural datasets and benchmarks are comparably rare. We present an\nannotated dataset and benchmarks for the semantic interpretation of real\nagricultural fields. Our dataset recorded with a UAV provides high-quality,\npixel-wise annotations of crops and weeds, but also crop leaf instances at the\nsame time. Furthermore, we provide benchmarks for various tasks on a hidden\ntest set comprised of different fields: known fields covered by the training\ndata and a completely unseen field. Our dataset, benchmarks, and code are\navailable at \\url{https://www.phenobench.org}.\n", "date": "2023-06-07", "categories": []}
{"id": "2407.16829", "title": "PlantTrack: Task-Driven Plant Keypoint Tracking with Zero-Shot Sim2Real\n  Transfer", "abstract": "  Tracking plant features is crucial for various agricultural tasks like\nphenotyping, pruning, or harvesting, but the unstructured, cluttered, and\ndeformable nature of plant environments makes it a challenging task. In this\ncontext, the recent advancements in foundational models show promise in\naddressing this challenge. In our work, we propose PlantTrack where we utilize\nDINOv2 which provides high-dimensional features, and train a keypoint heatmap\npredictor network to identify the locations of semantic features such as fruits\nand leaves which are then used as prompts for point tracking across video\nframes using TAPIR. We show that with as few as 20 synthetic images for\ntraining the keypoint predictor, we achieve zero-shot Sim2Real transfer,\nenabling effective tracking of plant features in real environments.\n", "date": "2024-07-23", "categories": []}
{"id": "2407.14737", "title": "Early Detection of Coffee Leaf Rust Through Convolutional Neural\n  Networks Trained on Low-Resolution Images", "abstract": "  Coffee leaf rust, a foliar disease caused by the fungus Hemileia vastatrix,\nposes a major threat to coffee production, especially in Central America.\nClimate change further aggravates this issue, as it shortens the latency period\nbetween initial infection and the emergence of visible symptoms in diseases\nlike leaf rust. Shortened latency periods can lead to more severe plant\nepidemics and faster spread of diseases. There is, hence, an urgent need for\neffective disease management strategies. To address these challenges, we\nexplore the potential of deep learning models for enhancing early disease\ndetection. However, deep learning models require extensive processing power and\nlarge amounts of data for model training, resources that are typically scarce.\nTo overcome these barriers, we propose a preprocessing technique that involves\nconvolving training images with a high-pass filter to enhance lesion-leaf\ncontrast, significantly improving model efficacy in resource-limited\nenvironments. This method and our model demonstrated a strong performance,\nachieving over 90% across all evaluation metrics--including precision, recall,\nF1-score, and the Dice coefficient. Our experiments show that this approach\noutperforms other methods, including two different image preprocessing\ntechniques and using unaltered, full-color images.\n", "date": "2024-07-20", "categories": []}
{"id": "2407.14119", "title": "Shape and Style GAN-based Multispectral Data Augmentation for Crop/Weed\n  Segmentation in Precision Farming", "abstract": "  The use of deep learning methods for precision farming is gaining increasing\ninterest. However, collecting training data in this application field is\nparticularly challenging and costly due to the need of acquiring information\nduring the different growing stages of the cultivation of interest. In this\npaper, we present a method for data augmentation that uses two GANs to create\nartificial images to augment the training data. To obtain a higher image\nquality, instead of re-creating the entire scene, we take original images and\nreplace only the patches containing objects of interest with artificial ones\ncontaining new objects with different shapes and styles. In doing this, we take\ninto account both the foreground (i.e., crop samples) and the background (i.e.,\nthe soil) of the patches. Quantitative experiments, conducted on publicly\navailable datasets, demonstrate the effectiveness of the proposed approach. The\nsource code and data discussed in this work are available as open source.\n", "date": "2024-07-19", "categories": []}
{"id": "2407.12614", "title": "Strawberry detection and counting based on YOLOv7 pruning and\n  information based tracking algorithm", "abstract": "  The strawberry industry yields significant economic benefits for Florida, yet\nthe process of monitoring strawberry growth and yield is labor-intensive and\ncostly. The development of machine learning-based detection and tracking\nmethodologies has been used for helping automated monitoring and prediction of\nstrawberry yield, still, enhancement has been limited as previous studies only\napplied the deep learning method for flower and fruit detection, which did not\nconsider the unique characteristics of image datasets collected by the machine\nvision system. This study proposed an optimal pruning of detection heads of the\ndeep learning model (YOLOv7 and its variants) that could achieve fast and\nprecise strawberry flower, immature fruit, and mature fruit detection.\nThereafter, an enhanced object tracking algorithm, which is called the\nInformation Based Tracking Algorithm (IBTA) utilized the best detection result,\nremoved the Kalman Filter, and integrated moving direction, velocity, and\nspatial information to improve the precision in strawberry flower and fruit\ntracking. The proposed pruning of detection heads across YOLOv7 variants,\nnotably Pruning-YOLOv7-tiny with detection head 3 and Pruning-YOLOv7-tiny with\nheads 2 and 3 achieved the best inference speed (163.9 frames per second) and\ndetection accuracy (89.1%), respectively. On the other hand, the effect of IBTA\nwas proved by comparing it with the centroid tracking algorithm (CTA), the\nMultiple Object Tracking Accuracy (MOTA) and Multiple Object Tracking Precision\n(MOTP) of IBTA were 12.3% and 6.0% higher than that of CTA, accordingly. In\naddition, other object-tracking evaluation metrics, including IDF1, IDR, IDP,\nMT, and IDs, show that IBTA performed better than CTA in strawberry flower and\nfruit tracking.\n", "date": "2024-07-17", "categories": []}
{"id": "2407.11275", "title": "M18K: A Comprehensive RGB-D Dataset and Benchmark for Mushroom Detection\n  and Instance Segmentation", "abstract": "  Automating agricultural processes holds significant promise for enhancing\nefficiency and sustainability in various farming practices. This paper\ncontributes to the automation of agricultural processes by providing a\ndedicated mushroom detection dataset related to automated harvesting, growth\nmonitoring, and quality control of the button mushroom produced using Agaricus\nBisporus fungus. With over 18,000 mushroom instances in 423 RGB-D image pairs\ntaken with an Intel RealSense D405 camera, it fills the gap in\nmushroom-specific datasets and serves as a benchmark for detection and instance\nsegmentation algorithms in smart mushroom agriculture. The dataset, featuring\nrealistic growth environment scenarios with comprehensive annotations, is\nassessed using advanced detection and instance segmentation algorithms. The\npaper details the dataset's characteristics, evaluates algorithmic performance,\nand for broader applicability, we have made all resources publicly available\nincluding images, codes, and trained models via our GitHub repository\nhttps://github.com/abdollahzakeri/m18k\n", "date": "2024-07-15", "categories": []}
{"id": "2407.11753", "title": "A Channel Attention-Driven Hybrid CNN Framework for Paddy Leaf Disease\n  Detection", "abstract": "  Farmers face various challenges when it comes to identifying diseases in rice\nleaves during their early stages of growth, which is a major reason for poor\nproduce. Therefore, early and accurate disease identification is important in\nagriculture to avoid crop loss and improve cultivation. In this research, we\npropose a novel hybrid deep learning (DL) classifier designed by extending the\nSqueeze-and-Excitation network architecture with a channel attention mechanism\nand the Swish ReLU activation function. The channel attention mechanism in our\nproposed model identifies the most important feature channels required for\nclassification during feature extraction and selection. The dying ReLU problem\nis mitigated by utilizing the Swish ReLU activation function, and the\nSqueeze-andExcitation blocks improve information propagation and cross-channel\ninteraction. Upon evaluation, our model achieved a high F1-score of 99.76% and\nan accuracy of 99.74%, surpassing the performance of existing models. These\noutcomes demonstrate the potential of state-of-the-art DL techniques in\nagriculture, contributing to the advancement of more efficient and reliable\ndisease detection systems.\n", "date": "2024-07-16", "categories": []}
{"id": "2407.10413", "title": "Melon Fruit Detection and Quality Assessment Using Generative AI-Based\n  Image Data Augmentation", "abstract": "  Monitoring and managing the growth and quality of fruits are very important\ntasks. To effectively train deep learning models like YOLO for real-time fruit\ndetection, high-quality image datasets are essential. However, such datasets\nare often lacking in agriculture. Generative AI models can help create\nhigh-quality images. In this study, we used MidJourney and Firefly tools to\ngenerate images of melon greenhouses and post-harvest fruits through\ntext-to-image, pre-harvest image-to-image, and post-harvest image-to-image\nmethods. We evaluated these AIgenerated images using PSNR and SSIM metrics and\ntested the detection performance of the YOLOv9 model. We also assessed the net\nquality of real and generated fruits. Our results showed that generative AI\ncould produce images very similar to real ones, especially for post-harvest\nfruits. The YOLOv9 model detected the generated images well, and the net\nquality was also measurable. This shows that generative AI can create realistic\nimages useful for fruit detection and quality assessment, indicating its great\npotential in agriculture. This study highlights the potential of AI-generated\nimages for data augmentation in melon fruit detection and quality assessment\nand envisions a positive future for generative AI applications in agriculture.\n", "date": "2024-07-15", "categories": []}
{"id": "2407.04711", "title": "MetaFruit Meets Foundation Models: Leveraging a Comprehensive\n  Multi-Fruit Dataset for Advancing Agricultural Foundation Models", "abstract": "  Fruit harvesting poses a significant labor and financial burden for the\nindustry, highlighting the critical need for advancements in robotic harvesting\nsolutions. Machine vision-based fruit detection has been recognized as a\ncrucial component for robust identification of fruits to guide robotic\nmanipulation. Despite considerable progress in leveraging deep learning and\nmachine learning techniques for fruit detection, a common shortfall is the\ninability to swiftly extend the developed models across different orchards\nand/or various fruit species. Additionally, the limited availability of\npertinent data further compounds these challenges. In this work, we introduce\nMetaFruit, the largest publicly available multi-class fruit dataset, comprising\n4,248 images and 248,015 manually labeled instances across diverse U.S.\norchards. Furthermore, this study proposes an innovative open-set fruit\ndetection system leveraging advanced Vision Foundation Models (VFMs) for fruit\ndetection that can adeptly identify a wide array of fruit types under varying\norchard conditions. This system not only demonstrates remarkable adaptability\nin learning from minimal data through few-shot learning but also shows the\nability to interpret human instructions for subtle detection tasks. The\nperformance of the developed foundation model is comprehensively evaluated\nusing several metrics, which outperforms the existing state-of-the-art\nalgorithms in both our MetaFruit dataset and other open-sourced fruit datasets,\nthereby setting a new benchmark in the field of agricultural technology and\nrobotic harvesting. The MetaFruit dataset and detection framework are\nopen-sourced to foster future research in vision-based fruit harvesting,\nmarking a significant stride toward addressing the urgent needs of the\nagricultural sector.\n", "date": "2024-05-14", "categories": []}
{"id": "2405.09118", "title": "BonnBot-I Plus: A Bio-diversity Aware Precise Weed Management Robotic\n  Platform", "abstract": "  In this article, we focus on the critical tasks of plant protection in arable\nfarms, addressing a modern challenge in agriculture: integrating ecological\nconsiderations into the operational strategy of precision weeding robots like\n\\bbot. This article presents the recent advancements in weed management\nalgorithms and the real-world performance of \\bbot\\ at the University of Bonn's\nKlein-Altendorf campus. We present a novel Rolling-view observation model for\nthe BonnBot-Is weed monitoring section which leads to an average absolute\nweeding performance enhancement of $3.4\\%$. Furthermore, for the first time, we\nshow how precision weeding robots could consider bio-diversity-aware concerns\nin challenging weeding scenarios. We carried out comprehensive weeding\nexperiments in sugar-beet fields, covering both weed-only and mixed crop-weed\nsituations, and introduced a new dataset compatible with precision weeding. Our\nreal-field experiments revealed that our weeding approach is capable of\nhandling diverse weed distributions, with a minimal loss of only $11.66\\%$\nattributable to intervention planning and $14.7\\%$ to vision system limitations\nhighlighting required improvements of the vision system.\n", "date": "2024-05-15", "categories": []}
{"id": "2312.09259", "title": "Livestock feeding behaviour: A review on automated systems for ruminant\n  monitoring", "abstract": "  Livestock feeding behaviour is an influential research area for those\ninvolved in animal husbandry and agriculture. In recent years, there has been a\ngrowing interest in automated systems for monitoring the behaviour of\nruminants. Despite the developments accomplished in the last decade, there is\nstill much to do and learn about the methods for measuring and analysing\nlivestock feeding behaviour. Automated monitoring systems mainly use motion,\nacoustic, and image sensors to collect animal behavioural data. The performance\nevaluation of existing methods is a complex task and direct comparisons between\nstudies are difficult. Several factors prevent a direct comparison, starting\nfrom the diversity of data and performance metrics used in the experiments. To\nthe best of our knowledge, this work represents the first tutorial-style review\non the analysis of the feeding behaviour of ruminants, emphasising the\nrelationship between sensing methodologies, signal processing, and\ncomputational intelligence methods. It assesses the main sensing methodologies\n(i.e. based on movement, sound, images/videos, and pressure) and the main\ntechniques to measure and analyse the signals associated with feeding\nbehaviour, evaluating their use in different settings and situations. It also\nhighlights the potentiality of automated monitoring systems to provide valuable\ninformation that improves our understanding of livestock feeding behaviour. The\nrelevance of these systems is increasingly important due to their impact on\nproduction systems and research. Finally, the paper closes by discussing future\nchallenges and opportunities in livestock feeding behaviour monitoring.\n", "date": "2023-12-03", "categories": []}
{"id": "2407.01435", "title": "Scarecrow monitoring system:employing mobilenet ssd for enhanced animal\n  supervision", "abstract": "  Agriculture faces a growing challenge with wildlife wreaking havoc on crops,\nthreatening sustainability. The project employs advanced object detection, the\nsystem utilizes the Mobile Net SSD model for real-time animal classification.\nThe methodology initiates with the creation of a dataset, where each animal is\nrepresented by annotated images. The SSD Mobile Net architecture facilitates\nthe use of a model for image classification and object detection. The model\nundergoes fine-tuning and optimization during training, enhancing accuracy for\nprecise animal classification. Real-time detection is achieved through a webcam\nand the OpenCV library, enabling prompt identification and categorization of\napproaching animals. By seamlessly integrating intelligent scarecrow technology\nwith object detection, this system offers a robust solution to field\nprotection, minimizing crop damage and promoting precision farming. It\nrepresents a valuable contribution to agricultural sustainability, addressing\nthe challenge of wildlife interference with crops. The implementation of the\nIntelligent Scarecrow Monitoring System stands as a progressive tool for\nproactive field management and protection, empowering farmers with an advanced\nsolution for precision agriculture.\n  Keywords: Machine learning, Deep Learning, Computer Vision, MobileNet SSD\n", "date": "2024-07-01", "categories": []}
{"id": "2406.12271", "title": "Agriculture-Vision Challenge 2024 -- The Runner-Up Solution for\n  Agricultural Pattern Recognition via Class Balancing and Model Ensemble", "abstract": "  The Agriculture-Vision Challenge at CVPR 2024 aims at leveraging semantic\nsegmentation models to produce pixel level semantic segmentation labels within\nregions of interest for multi-modality satellite images. It is one of the most\nfamous and competitive challenges for global researchers to break the boundary\nbetween computer vision and agriculture sectors. However, there is a serious\nclass imbalance problem in the agriculture-vision dataset, which hinders the\nsemantic segmentation performance. To solve this problem, firstly, we propose a\nmosaic data augmentation with a rare class sampling strategy to enrich\nlong-tail class samples. Secondly, we employ an adaptive class weight scheme to\nsuppress the contribution of the common classes while increasing the ones of\nrare classes. Thirdly, we propose a probability post-process to increase the\npredicted value of the rare classes. Our methodology achieved a mean\nIntersection over Union (mIoU) score of 0.547 on the test set, securing second\nplace in this challenge.\n", "date": "2024-06-18", "categories": []}
{"id": "2310.01055", "title": "Improved Crop and Weed Detection with Diverse Data Ensemble Learning", "abstract": "  Modern agriculture heavily relies on Site-Specific Farm Management practices,\nnecessitating accurate detection, localization, and quantification of crops and\nweeds in the field, which can be achieved using deep learning techniques. In\nthis regard, crop and weed-specific binary segmentation models have shown\npromise. However, uncontrolled field conditions limit their performance from\none field to the other. To improve semantic model generalization, existing\nmethods augment and synthesize agricultural data to account for uncontrolled\nfield conditions. However, given highly varied field conditions, these methods\nhave limitations. To overcome the challenges of model deterioration in such\nconditions, we propose utilizing data specific to other crops and weeds for our\nspecific target problem. To achieve this, we propose a novel ensemble\nframework. Our approach involves utilizing different crop and weed models\ntrained on diverse datasets and employing a teacher-student configuration. By\nusing homogeneous stacking of base models and a trainable meta-architecture to\ncombine their outputs, we achieve significant improvements for Canola crops and\nKochia weeds on unseen test data, surpassing the performance of single semantic\nsegmentation models. We identify the UNET meta-architecture as the most\neffective in this context. Finally, through ablation studies, we demonstrate\nand validate the effectiveness of our proposed model. We observe that including\nbase models trained on other target crops and weeds can help generalize the\nmodel to capture varied field conditions. Lastly, we propose two novel datasets\nwith varied conditions for comparisons.\n", "date": "2023-10-02", "categories": []}
{"id": "2406.10139", "title": "YOLOv1 to YOLOv10: A comprehensive review of YOLO variants and their\n  application in the agricultural domain", "abstract": "  This survey investigates the transformative potential of various YOLO\nvariants, from YOLOv1 to the state-of-the-art YOLOv10, in the context of\nagricultural advancements. The primary objective is to elucidate how these\ncutting-edge object detection models can re-energise and optimize diverse\naspects of agriculture, ranging from crop monitoring to livestock management.\nIt aims to achieve key objectives, including the identification of contemporary\nchallenges in agriculture, a detailed assessment of YOLO's incremental\nadvancements, and an exploration of its specific applications in agriculture.\nThis is one of the first surveys to include the latest YOLOv10, offering a\nfresh perspective on its implications for precision farming and sustainable\nagricultural practices in the era of Artificial Intelligence and automation.\nFurther, the survey undertakes a critical analysis of YOLO's performance,\nsynthesizes existing research, and projects future trends. By scrutinizing the\nunique capabilities packed in YOLO variants and their real-world applications,\nthis survey provides valuable insights into the evolving relationship between\nYOLO variants and agriculture. The findings contribute towards a nuanced\nunderstanding of the potential for precision farming and sustainable\nagricultural practices, marking a significant step forward in the integration\nof advanced object detection technologies within the agricultural sector.\n", "date": "2024-06-14", "categories": []}
{"id": "2406.07581", "title": "A novel method for identifying rice seed purity based on hybrid machine\n  learning algorithms", "abstract": "  In the grain industry, the identification of seed purity is a crucial task as\nit is an important factor in evaluating the quality of seeds. For rice seeds,\nthis property allows for the reduction of unexpected influences of other\nvarieties on rice yield, nutrient composition, and price. However, in practice,\nthey are often mixed with seeds from others. This study proposes a novel method\nfor automatically identifying the rice seed purity of a certain rice variety\nbased on hybrid machine learning algorithms. The main idea is to use deep\nlearning architectures for extracting important features from the raw data and\nthen use machine learning algorithms for classification. Several experiments\nare conducted following a practical implementation to evaluate the performance\nof the proposed model. The obtained results show that the novel method improves\nsignificantly the performance of existing methods. Thus, it can be applied to\ndesign effective identification systems for rice seed purity.\n", "date": "2024-06-09", "categories": []}
{"id": "2406.07716", "title": "Unleashing the Power of Transfer Learning Model for Sophisticated Insect\n  Detection: Revolutionizing Insect Classification", "abstract": "  The purpose of the Insect Detection System for Crop and Plant Health is to\nkeep an eye out for and identify insect infestations in farming areas. By\nutilizing cutting-edge technology like computer vision and machine learning,\nthe system seeks to identify hazardous insects early and accurately. This would\nenable prompt response to save crops and maintain optimal plant health. The\nMethod of this study includes Data Acquisition, Preprocessing, Data splitting,\nModel Implementation and Model evaluation. Different models like MobileNetV2,\nResNet152V2, Xecption, Custom CNN was used in this study. In order to\ncategorize insect photos, a Convolutional Neural Network (CNN) based on the\nResNet152V2 architecture is constructed and evaluated in this work. Achieving\n99% training accuracy and 97% testing accuracy, ResNet152V2 demonstrates\nsuperior performance among four implemented models. The results highlight its\npotential for real-world applications in insect classification and entomology\nstudies, emphasizing efficiency and accuracy. To ensure food security and\nsustain agricultural output globally, finding insects is crucial. Cutting-edge\ntechnology, such as ResNet152V2 models, greatly influence automating and\nimproving the accuracy of insect identification. Efficient insect detection not\nonly minimizes crop losses but also enhances agricultural productivity,\ncontributing to sustainable food production. This underscores the pivotal role\nof technology in addressing challenges related to global food security.\n", "date": "2024-06-11", "categories": []}
{"id": "2406.07482", "title": "Comparing Deep Learning Models for Rice Mapping in Bhutan Using High\n  Resolution Satellite Imagery", "abstract": "  The Bhutanese government is increasing its utilization of technological\napproaches such as including Remote Sensing-based knowledge in their\ndecision-making process. This study focuses on crop type and crop extent in\nParo, one of the top rice-yielding districts in Bhutan, and employs publicly\navailable NICFI high-resolution satellite imagery from Planet. Two Deep\nLearning (DL) approaches, point-based (DNN) and patch-based (U-Net), models\nwere used in conjunction with cloud-computing platforms. Three different models\nper DL approaches (DNN and U-Net) were trained: 1) RGBN channels from Planet;\n2) RGBN and elevation data (RGBNE); 3) RGBN and Sentinel-1 (S1) data (RGBNS),\nand RGBN with E and S1 data (RGBNES). From this comprehensive analysis, the\nU-Net displayed higher performance metrics across both model training and model\nvalidation efforts. Among the U-Net model sets, the RGBN, RGBNE, RGBNS, and\nRGBNES models had an F1-score of 0.8546, 0.8563, 0.8467, and 0.8500\nrespectively. An independent model evaluation was performed and found a high\nlevel of performance variation across all the metrics. For this independent\nmodel evaluation, the U-Net RGBN, RGBNE, RGBNES, and RGBN models displayed the\nF1-scores of 0.5935, 0.6154, 0.5882, and 0.6582, suggesting U-Net RGBNES as the\nbest model. The study shows that the DL approaches can predict rice. Also, DL\nmethods can be used with the survey-based approaches currently utilized by the\nBhutan Department of Agriculture. Further, this study demonstrated the usage of\nregional land cover products such as SERVIR's RLCMS as a weak label approach to\ncapture different strata addressing the class imbalance problem and improving\nthe sampling design for DL application. Finally, through preliminary model\ntesting and comparisons outlined it was shown that using additional features\nsuch as NDVI, EVI, and NDWI did not drastically improve model performance.\n", "date": "2024-06-11", "categories": []}
{"id": "2406.00313", "title": "From Seedling to Harvest: The GrowingSoy Dataset for Weed Detection in\n  Soy Crops via Instance Segmentation", "abstract": "  Deep learning, particularly Convolutional Neural Networks (CNNs), has gained\nsignificant attention for its effectiveness in computer vision, especially in\nagricultural tasks. Recent advancements in instance segmentation have improved\nimage classification accuracy. In this work, we introduce a comprehensive\ndataset for training neural networks to detect weeds and soy plants through\ninstance segmentation. Our dataset covers various stages of soy growth,\noffering a chronological perspective on weed invasion's impact, with 1,000\nmeticulously annotated images. We also provide 6 state of the art models,\ntrained in this dataset, that can understand and detect soy and weed in every\nstage of the plantation process. By using this dataset for weed and soy\nsegmentation, we achieved a segmentation average precision of 79.1% and an\naverage recall of 69.2% across all plant classes, with the YOLOv8X model.\nMoreover, the YOLOv8M model attained 78.7% mean average precision (mAp-50) in\ncaruru weed segmentation, 69.7% in grassy weed segmentation, and 90.1% in soy\nplant segmentation.\n", "date": "2024-06-01", "categories": []}
{"id": "2406.01869", "title": "Fruit Classification System with Deep Learning and Neural Architecture\n  Search", "abstract": "  The fruit identification process involves analyzing and categorizing\ndifferent types of fruits based on their visual characteristics. This activity\ncan be achieved using a range of methodologies, encompassing manual\nexamination, conventional computer vision methodologies, and more sophisticated\nmethodologies employing machine learning and deep learning. Our study\nidentified a total of 15 distinct categories of fruit, consisting of class\nAvocado, Banana, Cherry, Apple Braeburn, Apple golden 1, Apricot, Grape, Kiwi,\nMango, Orange, Papaya, Peach, Pineapple, Pomegranate and Strawberry. Neural\nArchitecture Search (NAS) is a technological advancement employed within the\nrealm of deep learning and artificial intelligence, to automate conceptualizing\nand refining neural network topologies. NAS aims to identify neural network\nstructures that are highly suitable for tasks, such as the detection of fruits.\nOur suggested model with 99.98% mAP increased the detection performance of the\npreceding research study that used Fruit datasets. In addition, after the\ncompletion of the study, a comparative analysis was carried out to assess the\nfindings in conjunction with those of another research that is connected to the\ntopic. When compared to the findings of earlier studies, the detector that was\nproposed exhibited higher performance in terms of both its accuracy and its\nprecision.\n", "date": "2024-06-04", "categories": []}
{"id": "2405.21056", "title": "An Organic Weed Control Prototype using Directed Energy and Deep\n  Learning", "abstract": "  Organic weed control is a vital to improve crop yield with a sustainable\napproach. In this work, a directed energy weed control robot prototype\nspecifically designed for organic farms is proposed. The robot uses a novel\ndistributed array robot (DAR) unit for weed treatment. Soybean and corn\ndatabases are built to train deep learning neural nets to perform weed\nrecognition. The initial deep learning neural nets show a high performance in\nclassifying crops. The robot uses a patented directed energy plant eradication\nrecipe that is completely organic and UV-C free, with no chemical damage or\nphysical disturbance to the soil. The deep learning can classify 8 common weed\nspecies in a soybean field under natural environment with up to 98% accuracy.\n", "date": "2024-05-31", "categories": []}
{"id": "2405.19413", "title": "VisTA-SR: Improving the Accuracy and Resolution of Low-Cost Thermal\n  Imaging Cameras for Agriculture", "abstract": "  Thermal cameras are an important tool for agricultural research because they\nallow for non-invasive measurement of plant temperature, which relates to\nimportant photochemical, hydraulic, and agronomic traits. Utilizing low-cost\nthermal cameras can lower the barrier to introducing thermal imaging in\nagricultural research and production. This paper presents an approach to\nimprove the temperature accuracy and image quality of low-cost thermal imaging\ncameras for agricultural applications. Leveraging advancements in computer\nvision techniques, particularly deep learning networks, we propose a method,\ncalled $\\textbf{VisTA-SR}$ ($\\textbf{Vis}$ual \\& $\\textbf{T}$hermal\n$\\textbf{A}$lignment and $\\textbf{S}$uper-$\\textbf{R}$esolution Enhancement)\nthat combines RGB and thermal images to enhance the capabilities of\nlow-resolution thermal cameras. The research includes calibration and\nvalidation of temperature measurements, acquisition of paired image datasets,\nand the development of a deep learning network tailored for agricultural\nthermal imaging. Our study addresses the challenges of image enhancement in the\nagricultural domain and explores the potential of low-cost thermal cameras to\nreplace high-resolution industrial cameras. Experimental results demonstrate\nthe effectiveness of our approach in enhancing temperature accuracy and image\nsharpness, paving the way for more accessible and efficient thermal imaging\nsolutions in agriculture.\n", "date": "2024-05-29", "categories": []}
{"id": "2309.11989", "title": "A Vision-Based Navigation System for Arable Fields", "abstract": "  Vision-based navigation systems in arable fields are an underexplored area in\nagricultural robot navigation. Vision systems deployed in arable fields face\nchallenges such as fluctuating weed density, varying illumination levels,\ngrowth stages and crop row irregularities. Current solutions are often\ncrop-specific and aimed to address limited individual conditions such as\nillumination or weed density. Moreover, the scarcity of comprehensive datasets\nhinders the development of generalised machine learning systems for navigating\nthese fields. This paper proposes a suite of deep learning-based perception\nalgorithms using affordable vision sensors for vision-based navigation in\narable fields. Initially, a comprehensive dataset that captures the intricacies\nof multiple crop seasons, various crop types, and a range of field variations\nwas compiled. Next, this study delves into the creation of robust infield\nperception algorithms capable of accurately detecting crop rows under diverse\nconditions such as different growth stages, weed density, and varying\nillumination. Further, it investigates the integration of crop row following\nwith vision-based crop row switching for efficient field-scale navigation. The\nproposed infield navigation system was tested in commercial arable fields\ntraversing a total distance of 4.5 km with average heading and cross-track\nerrors of 1.24{\\deg} and 3.32 cm respectively.\n", "date": "2023-09-21", "categories": []}
{"id": "2402.15909", "title": "Enhanced Droplet Analysis Using Generative Adversarial Networks", "abstract": "  Precision devices play an important role in enhancing production quality and\nproductivity in agricultural systems. Therefore, the optimization of these\ndevices is essential in precision agriculture. Recently, with the advancements\nof deep learning, there have been several studies aiming to harness its\ncapabilities for improving spray system performance. However, the effectiveness\nof these methods heavily depends on the size of the training dataset, which is\nexpensive and time-consuming to collect. To address the challenge of\ninsufficient training samples, we developed an image generator named DropletGAN\nto generate images of droplets. The DropletGAN model is trained by using a\nsmall dataset captured by a high-speed camera and capable of generating images\nwith progressively increasing resolution. The results demonstrate that the\nmodel can generate high-quality images with the size of 1024x1024. The\ngenerated images from the DropletGAN are evaluated using the Fr\\'echet\ninception distance (FID) with an FID score of 11.29. Furthermore, this research\nleverages recent advancements in computer vision and deep learning to develop a\nlight droplet detector using the synthetic dataset. As a result, the detection\nmodel achieves a 16.06% increase in mean average precision (mAP) when utilizing\nthe synthetic dataset. To the best of our knowledge, this work stands as the\nfirst to employ a generative model for augmenting droplet detection. Its\nsignificance lies not only in optimizing nozzle design for constructing\nefficient spray systems but also in addressing the common challenge of\ninsufficient data in various precision agriculture tasks. This work offers a\ncritical contribution to conserving resources while striving for optimal and\nsustainable agricultural practices.\n", "date": "2024-02-24", "categories": []}
{"id": "2405.11351", "title": "PlantTracing: Tracing Arabidopsis Thaliana Apex with CenterTrack", "abstract": "  This work applies an encoder-decoder-based machine learning network to detect\nand track the motion and growth of the flowering stem apex of Arabidopsis\nThaliana. Based on the CenterTrack, a machine learning back-end network, we\ntrained a model based on ten time-lapsed labeled videos and tested against\nthree videos.\n", "date": "2024-05-18", "categories": []}
{"id": "2405.10041", "title": "Revealing Hierarchical Structure of Leaf Venations in Plant Science via\n  Label-Efficient Segmentation: Dataset and Method", "abstract": "  Hierarchical leaf vein segmentation is a crucial but under-explored task in\nagricultural sciences, where analysis of the hierarchical structure of plant\nleaf venation can contribute to plant breeding. While current segmentation\ntechniques rely on data-driven models, there is no publicly available dataset\nspecifically designed for hierarchical leaf vein segmentation. To address this\ngap, we introduce the HierArchical Leaf Vein Segmentation (HALVS) dataset, the\nfirst public hierarchical leaf vein segmentation dataset. HALVS comprises 5,057\nreal-scanned high-resolution leaf images collected from three plant species:\nsoybean, sweet cherry, and London planetree. It also includes human-annotated\nground truth for three orders of leaf veins, with a total labeling effort of\n83.8 person-days. Based on HALVS, we further develop a label-efficient learning\nparadigm that leverages partial label information, i.e. missing annotations for\ntertiary veins. Empirical studies are performed on HALVS, revealing new\nobservations, challenges, and research directions on leaf vein segmentation.\n", "date": "2024-05-16", "categories": []}
{"id": "2311.10365", "title": "Dates Fruit Disease Recognition using Machine Learning", "abstract": "  Many countries such as Saudi Arabia, Morocco and Tunisia are among the top\nexporters and consumers of palm date fruits. Date fruit production plays a\nmajor role in the economies of the date fruit exporting countries. Date fruits\nare susceptible to disease just like any fruit and early detection and\nintervention can end up saving the produce. However, with the vast farming\nlands, it is nearly impossible for farmers to observe date trees on a frequent\nbasis for early disease detection. In addition, even with human observation the\nprocess is prone to human error and increases the date fruit cost. With the\nrecent advances in computer vision, machine learning, drone technology, and\nother technologies; an integrated solution can be proposed for the automatic\ndetection of date fruit disease. In this paper, a hybrid features based method\nwith the standard classifiers is proposed based on the extraction of L*a*b\ncolor features, statistical features, and Discrete Wavelet Transform (DWT)\ntexture features for the early detection and classification of date fruit\ndisease. A dataset was developed for this work consisting of 871 images divided\ninto the following classes; Healthy date, Initial stage of disease,\nMalnourished date, and Parasite infected. The extracted features were input to\ncommon classifiers such as the Random Forest (RF), Multilayer Perceptron (MLP),\nNa\\\"ive Bayes (NB), and Fuzzy Decision Trees (FDT). The highest average\naccuracy was achieved when combining the L*a*b, Statistical, and DWT Features.\n", "date": "2023-11-17", "categories": []}
{"id": "2405.07157", "title": "Semi-Self-Supervised Domain Adaptation: Developing Deep Learning Models\n  with Limited Annotated Data for Wheat Head Segmentation", "abstract": "  Precision agriculture involves the application of advanced technologies to\nimprove agricultural productivity, efficiency, and profitability while\nminimizing waste and environmental impact. Deep learning approaches enable\nautomated decision-making for many visual tasks. However, in the agricultural\ndomain, variability in growth stages and environmental conditions, such as\nweather and lighting, presents significant challenges to developing deep\nlearning-based techniques that generalize across different conditions. The\nresource-intensive nature of creating extensive annotated datasets that capture\nthese variabilities further hinders the widespread adoption of these\napproaches. To tackle these issues, we introduce a semi-self-supervised domain\nadaptation technique based on deep convolutional neural networks with a\nprobabilistic diffusion process, requiring minimal manual data annotation.\nUsing only three manually annotated images and a selection of video clips from\nwheat fields, we generated a large-scale computationally annotated dataset of\nimage-mask pairs and a large dataset of unannotated images extracted from video\nframes. We developed a two-branch convolutional encoder-decoder model\narchitecture that uses both synthesized image-mask pairs and unannotated\nimages, enabling effective adaptation to real images. The proposed model\nachieved a Dice score of 80.7\\% on an internal test dataset and a Dice score of\n64.8\\% on an external test set, composed of images from five countries and\nspanning 18 domains, indicating its potential to develop generalizable\nsolutions that could encourage the wider adoption of advanced technologies in\nagriculture.\n", "date": "2024-05-12", "categories": []}
{"id": "2405.07349", "title": "WeedScout: Real-Time Autonomous blackgrass Classification and Mapping\n  using dedicated hardware", "abstract": "  Blackgrass (Alopecurus myosuroides) is a competitive weed that has\nwide-ranging impacts on food security by reducing crop yields and increasing\ncultivation costs. In addition to the financial burden on agriculture, the\napplication of herbicides as a preventive to blackgrass can negatively affect\naccess to clean water and sanitation. The WeedScout project introduces a\nReal-Rime Autonomous Black-Grass Classification and Mapping (RT-ABGCM), a\ncutting-edge solution tailored for real-time detection of blackgrass, for\nprecision weed management practices. Leveraging Artificial Intelligence (AI)\nalgorithms, the system processes live image feeds, infers blackgrass density,\nand covers two stages of maturation. The research investigates the deployment\nof You Only Look Once (YOLO) models, specifically the streamlined YOLOv8 and\nYOLO-NAS, accelerated at the edge with the NVIDIA Jetson Nano (NJN). By\noptimising inference speed and model performance, the project advances the\nintegration of AI into agricultural practices, offering potential solutions to\nchallenges such as herbicide resistance and environmental impact. Additionally,\ntwo datasets and model weights are made available to the research community,\nfacilitating further advancements in weed detection and precision farming\ntechnologies.\n", "date": "2024-05-12", "categories": []}
{"id": "2405.07399", "title": "Semi-Supervised Weed Detection for Rapid Deployment and Enhanced\n  Efficiency", "abstract": "  Weeds present a significant challenge in agriculture, causing yield loss and\nrequiring expensive control measures. Automatic weed detection using computer\nvision and deep learning offers a promising solution. However, conventional\ndeep learning methods often require large amounts of labelled training data,\nwhich can be costly and time-consuming to acquire. This paper introduces a\nnovel method for semi-supervised weed detection, comprising two main\ncomponents. Firstly, a multi-scale feature representation technique is employed\nto capture distinctive weed features across different scales. Secondly, we\npropose an adaptive pseudo-label assignment strategy, leveraging a small set of\nlabelled images during training. This strategy dynamically assigns confidence\nscores to pseudo-labels generated from unlabeled data. Additionally, our\napproach integrates epoch-corresponding and mixed pseudo-labels to further\nenhance the learning process. Experimental results on the COCO dataset and five\nprominent weed datasets -- CottonWeedDet12, CropAndWeed, Palmer amaranth,\nRadishWheat, and RoboWeedMap -- illustrate that our method achieves\nstate-of-the-art performance in weed detection, even with significantly less\nlabelled data compared to existing techniques. This approach holds the\npotential to alleviate the labelling burden and enhance the feasibility and\ndeployment speed of deep learning for weed detection in real-world agricultural\nscenarios.\n", "date": "2024-05-12", "categories": []}
{"id": "2405.06260", "title": "Precise Apple Detection and Localization in Orchards using YOLOv5 for\n  Robotic Harvesting Systems", "abstract": "  The advancement of agricultural robotics holds immense promise for\ntransforming fruit harvesting practices, particularly within the apple\nindustry. The accurate detection and localization of fruits are pivotal for the\nsuccessful implementation of robotic harvesting systems. In this paper, we\npropose a novel approach to apple detection and position estimation utilizing\nan object detection model, YOLOv5. Our primary objective is to develop a robust\nsystem capable of identifying apples in complex orchard environments and\nproviding precise location information. To achieve this, we curated an\nautonomously labeled dataset comprising diverse apple tree images, which was\nutilized for both training and evaluation purposes. Through rigorous\nexperimentation, we compared the performance of our YOLOv5-based system with\nother popular object detection models, including SSD. Our results demonstrate\nthat the YOLOv5 model outperforms its counterparts, achieving an impressive\napple detection accuracy of approximately 85%. We believe that our proposed\nsystem's accurate apple detection and position estimation capabilities\nrepresent a significant advancement in agricultural robotics, laying the\ngroundwork for more efficient and sustainable fruit harvesting practices.\n", "date": "2024-05-10", "categories": []}
{"id": "2405.03702", "title": "Leafy Spurge Dataset: Real-world Weed Classification Within Aerial Drone\n  Imagery", "abstract": "  Invasive plant species are detrimental to the ecology of both agricultural\nand wildland areas. Euphorbia esula, or leafy spurge, is one such plant that\nhas spread through much of North America from Eastern Europe. When paired with\ncontemporary computer vision systems, unmanned aerial vehicles, or drones,\noffer the means to track expansion of problem plants, such as leafy spurge, and\nimprove chances of controlling these weeds. We gathered a dataset of leafy\nspurge presence and absence in grasslands of western Montana, USA, then\nsurveyed these areas with a commercial drone. We trained image classifiers on\nthese data, and our best performing model, a pre-trained DINOv2 vision\ntransformer, identified leafy spurge with 0.84 accuracy (test set). This result\nindicates that classification of leafy spurge is tractable, but not solved. We\nrelease this unique dataset of labelled and unlabelled, aerial drone imagery\nfor the machine learning community to explore. Improving classification\nperformance of leafy spurge would benefit the fields of ecology, conservation,\nand remote sensing alike. Code and data are available at our website:\nleafy-spurge-dataset.github.io.\n", "date": "2024-05-02", "categories": []}
{"id": "2405.04535", "title": "Image Classification for CSSVD Detection in Cacao Plants", "abstract": "  The detection of diseases within plants has attracted a lot of attention from\ncomputer vision enthusiasts. Despite the progress made to detect diseases in\nmany plants, there remains a research gap to train image classifiers to detect\nthe cacao swollen shoot virus disease or CSSVD for short, pertinent to cacao\nplants. This gap has mainly been due to the unavailability of high quality\nlabeled training data. Moreover, institutions have been hesitant to share their\ndata related to CSSVD. To fill these gaps, we propose the development of image\nclassifiers to detect CSSVD-infected cacao plants. Our proposed solution is\nbased on VGG16, ResNet50 and Vision Transformer (ViT). We evaluate the\nclassifiers on a recently released and publicly accessible KaraAgroAI Cocoa\ndataset. Our best image classifier, based on ResNet50, achieves 95.39\\%\nprecision, 93.75\\% recall, 94.34\\% F1-score and 94\\% accuracy on only 20\nepochs. There is a +9.75\\% improvement in recall when compared to previous\nworks. Our results indicate that the image classifiers learn to identify cacao\nplants infected with CSSVD.\n", "date": "2024-03-13", "categories": []}
{"id": "2405.04305", "title": "A New Dataset and Comparative Study for Aphid Cluster Detection and\n  Segmentation in Sorghum Fields", "abstract": "  Aphid infestations are one of the primary causes of extensive damage to wheat\nand sorghum fields and are one of the most common vectors for plant viruses,\nresulting in significant agricultural yield losses. To address this problem,\nfarmers often employ the inefficient use of harmful chemical pesticides that\nhave negative health and environmental impacts. As a result, a large amount of\npesticide is wasted on areas without significant pest infestation. This brings\nto attention the urgent need for an intelligent autonomous system that can\nlocate and spray sufficiently large infestations selectively within the complex\ncrop canopies. We have developed a large multi-scale dataset for aphid cluster\ndetection and segmentation, collected from actual sorghum fields and\nmeticulously annotated to include clusters of aphids. Our dataset comprises a\ntotal of 54,742 image patches, showcasing a variety of viewpoints, diverse\nlighting conditions, and multiple scales, highlighting its effectiveness for\nreal-world applications. In this study, we trained and evaluated four real-time\nsemantic segmentation models and three object detection models specifically for\naphid cluster segmentation and detection. Considering the balance between\naccuracy and efficiency, Fast-SCNN delivered the most effective segmentation\nresults, achieving 80.46% mean precision, 81.21% mean recall, and 91.66 frames\nper second (FPS). For object detection, RT-DETR exhibited the best overall\nperformance with a 61.63% mean average precision (mAP), 92.6% mean recall, and\n72.55 on an NVIDIA V100 GPU. Our experiments further indicate that aphid\ncluster segmentation is more suitable for assessing aphid infestations than\nusing detection models.\n", "date": "2024-05-07", "categories": []}
{"id": "2404.19748", "title": "Quantifying Nematodes through Images: Datasets, Models, and Baselines of\n  Deep Learning", "abstract": "  Every year, plant parasitic nematodes, one of the major groups of plant\npathogens, cause a significant loss of crops worldwide. To mitigate crop yield\nlosses caused by nematodes, an efficient nematode monitoring method is\nessential for plant and crop disease management. In other respects, efficient\nnematode detection contributes to medical research and drug discovery, as\nnematodes are model organisms. With the rapid development of computer\ntechnology, computer vision techniques provide a feasible solution for\nquantifying nematodes or nematode infections. In this paper, we survey and\ncategorise the studies and available datasets on nematode detection through\ndeep-learning models. To stimulate progress in related research, this survey\npresents the potential state-of-the-art object detection models, training\ntechniques, optimisation techniques, and evaluation metrics for deep learning\nbeginners. Moreover, seven state-of-the-art object detection models are\nvalidated on three public datasets and the AgriNema dataset for plant parasitic\nnematodes to construct a baseline for nematode detection.\n", "date": "2024-04-30", "categories": []}
{"id": "2404.17718", "title": "Lessons from Deploying CropFollow++: Under-Canopy Agricultural\n  Navigation with Keypoints", "abstract": "  We present a vision-based navigation system for under-canopy agricultural\nrobots using semantic keypoints. Autonomous under-canopy navigation is\nchallenging due to the tight spacing between the crop rows ($\\sim 0.75$ m),\ndegradation in RTK-GPS accuracy due to multipath error, and noise in LiDAR\nmeasurements from the excessive clutter. Our system, CropFollow++, introduces\nmodular and interpretable perception architecture with a learned semantic\nkeypoint representation. We deployed CropFollow++ in multiple under-canopy\ncover crop planting robots on a large scale (25 km in total) in various field\nconditions and we discuss the key lessons learned from this.\n", "date": "2024-04-26", "categories": []}
{"id": "2404.13996", "title": "Challenges in automatic and selective plant-clearing", "abstract": "  With the advent of multispectral imagery and AI, there have been numerous\nworks on automatic plant segmentation for purposes such as counting, picking,\nhealth monitoring, localized pesticide delivery, etc. In this paper, we tackle\nthe related problem of automatic and selective plant-clearing in a sustainable\nforestry context, where an autonomous machine has to detect and avoid specific\nplants while clearing any weeds which may compete with the species being\ncultivated. Such an autonomous system requires a high level of robustness to\nweather conditions, plant variability, terrain and weeds while remaining cheap\nand easy to maintain. We notably discuss the lack of robustness of spectral\nimagery, investigate the impact of the reference database's size and discuss\nissues specific to AI systems operating in uncontrolled environments.\n", "date": "2024-04-22", "categories": []}
{"id": "2404.11762", "title": "IrrNet: Advancing Irrigation Mapping with Incremental Patch Size\n  Training on Remote Sensing Imagery", "abstract": "  Irrigation mapping plays a crucial role in effective water management,\nessential for preserving both water quality and quantity, and is key to\nmitigating the global issue of water scarcity. The complexity of agricultural\nfields, adorned with diverse irrigation practices, especially when multiple\nsystems coexist in close quarters, poses a unique challenge. This complexity is\nfurther compounded by the nature of Landsat's remote sensing data, where each\npixel is rich with densely packed information, complicating the task of\naccurate irrigation mapping. In this study, we introduce an innovative approach\nthat employs a progressive training method, which strategically increases patch\nsizes throughout the training process, utilizing datasets from Landsat 5 and 7,\nlabeled with the WRLU dataset for precise labeling. This initial focus allows\nthe model to capture detailed features, progressively shifting to broader, more\ngeneral features as the patch size enlarges. Remarkably, our method enhances\nthe performance of existing state-of-the-art models by approximately 20%.\nFurthermore, our analysis delves into the significance of incorporating various\nspectral bands into the model, assessing their impact on performance. The\nfindings reveal that additional bands are instrumental in enabling the model to\ndiscern finer details more effectively. This work sets a new standard for\nleveraging remote sensing imagery in irrigation mapping.\n", "date": "2024-04-17", "categories": []}
{"id": "2404.08931", "title": "Label-free Anomaly Detection in Aerial Agricultural Images with Masked\n  Image Modeling", "abstract": "  Detecting various types of stresses (nutritional, water, nitrogen, etc.) in\nagricultural fields is critical for farmers to ensure maximum productivity.\nHowever, stresses show up in different shapes and sizes across different crop\ntypes and varieties. Hence, this is posed as an anomaly detection task in\nagricultural images. Accurate anomaly detection in agricultural UAV images is\nvital for early identification of field irregularities. Traditional supervised\nlearning faces challenges in adapting to diverse anomalies, necessitating\nextensive annotated data. In this work, we overcome this limitation with\nself-supervised learning using a masked image modeling approach. Masked\nAutoencoders (MAE) extract meaningful normal features from unlabeled image\nsamples which produces high reconstruction error for the abnormal pixels during\nreconstruction. To remove the need of using only ``normal\" data while training,\nwe use an anomaly suppression loss mechanism that effectively minimizes the\nreconstruction of anomalous pixels and allows the model to learn anomalous\nareas without explicitly separating ``normal\" images for training. Evaluation\non the Agriculture-Vision data challenge shows a mIOU score improvement in\ncomparison to prior state of the art in unsupervised and self-supervised\nmethods. A single model generalizes across all the anomaly categories in the\nAgri-Vision Challenge Dataset\n", "date": "2024-04-13", "categories": []}
{"id": "2404.08515", "title": "ChatGPT and general-purpose AI count fruits in pictures surprisingly\n  well", "abstract": "  Object counting is a popular task in deep learning applications in various\ndomains, including agriculture. A conventional deep learning approach requires\na large amount of training data, often a logistic problem in a real-world\napplication. To address this issue, we examined how well ChatGPT (GPT4V) and a\ngeneral-purpose AI (foundation model for object counting, T-Rex) can count the\nnumber of fruit bodies (coffee cherries) in 100 images. The foundation model\nwith few-shot learning outperformed the trained YOLOv8 model (R2 = 0.923 and\n0.900, respectively). ChatGPT also showed some interesting potential,\nespecially when few-shot learning with human feedback was applied (R2 = 0.360\nand 0.460, respectively). Moreover, we examined the time required for\nimplementation as a practical question. Obtaining the results with the\nfoundation model and ChatGPT were much shorter than the YOLOv8 model (0.83 hrs,\n1.75 hrs, and 161 hrs). We interpret these results as two surprises for deep\nlearning users in applied domains: a foundation model with few-shot\ndomain-specific learning can drastically save time and effort compared to the\nconventional approach, and ChatGPT can reveal a relatively good performance.\nBoth approaches do not need coding skills, which can foster AI education and\ndissemination.\n", "date": "2024-04-12", "categories": []}
{"id": "2404.03611", "title": "InsectMamba: Insect Pest Classification with State Space Model", "abstract": "  The classification of insect pests is a critical task in agricultural\ntechnology, vital for ensuring food security and environmental sustainability.\nHowever, the complexity of pest identification, due to factors like high\ncamouflage and species diversity, poses significant obstacles. Existing methods\nstruggle with the fine-grained feature extraction needed to distinguish between\nclosely related pest species. Although recent advancements have utilized\nmodified network structures and combined deep learning approaches to improve\naccuracy, challenges persist due to the similarity between pests and their\nsurroundings. To address this problem, we introduce InsectMamba, a novel\napproach that integrates State Space Models (SSMs), Convolutional Neural\nNetworks (CNNs), Multi-Head Self-Attention mechanism (MSA), and Multilayer\nPerceptrons (MLPs) within Mix-SSM blocks. This integration facilitates the\nextraction of comprehensive visual features by leveraging the strengths of each\nencoding strategy. A selective module is also proposed to adaptively aggregate\nthese features, enhancing the model's ability to discern pest characteristics.\nInsectMamba was evaluated against strong competitors across five insect pest\nclassification datasets. The results demonstrate its superior performance and\nverify the significance of each model component by an ablation study.\n", "date": "2024-04-04", "categories": []}
{"id": "2404.00364", "title": "Accurate Cutting-point Estimation for Robotic Lychee Harvesting through\n  Geometry-aware Learning", "abstract": "  Accurately identifying lychee-picking points in unstructured orchard\nenvironments and obtaining their coordinate locations is critical to the\nsuccess of lychee-picking robots. However, traditional two-dimensional (2D)\nimage-based object detection methods often struggle due to the complex\ngeometric structures of branches, leaves and fruits, leading to incorrect\ndetermination of lychee picking points. In this study, we propose a\nFcaf3d-lychee network model specifically designed for the accurate localisation\nof lychee picking points. Point cloud data of lychee picking points in natural\nenvironments are acquired using Microsoft's Azure Kinect DK time-of-flight\n(TOF) camera through multi-view stitching. We augment the Fully Convolutional\nAnchor-Free 3D Object Detection (Fcaf3d) model with a\nsqueeze-and-excitation(SE) module, which exploits human visual attention\nmechanisms for improved feature extraction of lychee picking points. The\ntrained network model is evaluated on a test set of lychee-picking locations\nand achieves an impressive F1 score of 88.57%, significantly outperforming\nexisting models. Subsequent three-dimensional (3D) position detection of\npicking points in real lychee orchard environments yields high accuracy, even\nunder varying degrees of occlusion. Localisation errors of lychee picking\npoints are within 1.5 cm in all directions, demonstrating the robustness and\ngenerality of the model.\n", "date": "2024-03-30", "categories": []}
{"id": "2404.00593", "title": "LAESI: Leaf Area Estimation with Synthetic Imagery", "abstract": "  We introduce LAESI, a Synthetic Leaf Dataset of 100,000 synthetic leaf images\non millimeter paper, each with semantic masks and surface area labels. This\ndataset provides a resource for leaf morphology analysis primarily aimed at\nbeech and oak leaves. We evaluate the applicability of the dataset by training\nmachine learning models for leaf surface area prediction and semantic\nsegmentation, using real images for validation. Our validation shows that these\nmodels can be trained to predict leaf surface area with a relative error not\ngreater than an average human annotator. LAESI also provides an efficient\nframework based on 3D procedural models and generative AI for the large-scale,\ncontrollable generation of data with potential further applications in\nagriculture and biology. We evaluate the inclusion of generative AI in our\nprocedural data generation pipeline and show how data filtering based on\nannotation consistency results in datasets which allow training the highest\nperforming vision models.\n", "date": "2024-03-31", "categories": []}
{"id": "2403.15981", "title": "Exploring Accurate 3D Phenotyping in Greenhouse through Neural Radiance\n  Fields", "abstract": "  Accurate collection of plant phenotyping is critical to optimising\nsustainable farming practices in precision agriculture. Traditional phenotyping\nin controlled laboratory environments, while valuable, falls short in\nunderstanding plant growth under real-world conditions. Emerging sensor and\ndigital technologies offer a promising approach for direct phenotyping of\nplants in farm environments. This study investigates a learning-based\nphenotyping method using the Neural Radiance Field to achieve accurate in-situ\nphenotyping of pepper plants in greenhouse environments. To quantitatively\nevaluate the performance of this method, traditional point cloud registration\non 3D scanning data is implemented for comparison. Experimental result shows\nthat NeRF(Neural Radiance Fields) achieves competitive accuracy compared to the\n3D scanning methods. The mean distance error between the scanner-based method\nand the NeRF-based method is 0.865mm. This study shows that the learning-based\nNeRF method achieves similar accuracy to 3D scanning-based methods but with\nimproved scalability and robustness.\n", "date": "2024-03-24", "categories": []}
{"id": "2403.15248", "title": "Self-Supervised Backbone Framework for Diverse Agricultural Vision Tasks", "abstract": "  Computer vision in agriculture is game-changing with its ability to transform\nfarming into a data-driven, precise, and sustainable industry. Deep learning\nhas empowered agriculture vision to analyze vast, complex visual data, but\nheavily rely on the availability of large annotated datasets. This remains a\nbottleneck as manual labeling is error-prone, time-consuming, and expensive.\nThe lack of efficient labeling approaches inspired us to consider\nself-supervised learning as a paradigm shift, learning meaningful feature\nrepresentations from raw agricultural image data. In this work, we explore how\nself-supervised representation learning unlocks the potential applicability to\ndiverse agriculture vision tasks by eliminating the need for large-scale\nannotated datasets. We propose a lightweight framework utilizing SimCLR, a\ncontrastive learning approach, to pre-train a ResNet-50 backbone on a large,\nunannotated dataset of real-world agriculture field images. Our experimental\nanalysis and results indicate that the model learns robust features applicable\nto a broad range of downstream agriculture tasks discussed in the paper.\nAdditionally, the reduced reliance on annotated data makes our approach more\ncost-effective and accessible, paving the way for broader adoption of computer\nvision in agriculture.\n", "date": "2024-03-22", "categories": []}
{"id": "2403.14359", "title": "Varroa destructor detection on honey bees using hyperspectral imagery", "abstract": "  Hyperspectral (HS) imagery in agriculture is becoming increasingly common.\nThese images have the advantage of higher spectral resolution. Advanced\nspectral processing techniques are required to unlock the information potential\nin these HS images. The present paper introduces a method rooted in\nmultivariate statistics designed to detect parasitic Varroa destructor mites on\nthe body of western honey bee Apis mellifera, enabling easier and continuous\nmonitoring of the bee hives. The methodology explores unsupervised (K-means++)\nand recently developed supervised (Kernel Flows - Partial Least-Squares,\nKF-PLS) methods for parasitic identification. Additionally, in light of the\nemergence of custom-band multispectral cameras, the present research outlines a\nstrategy for identifying the specific wavelengths necessary for effective\nbee-mite separation, suitable for implementation in a custom-band camera.\nIllustrated with a real-case dataset, our findings demonstrate that as few as\nfour spectral bands are sufficient for accurate parasite identification.\n", "date": "2024-03-21", "categories": []}
{"id": "2403.12935", "title": "Segment Anything for comprehensive analysis of grapevine cluster\n  architecture and berry properties", "abstract": "  Grape cluster architecture and compactness are complex traits influencing\ndisease susceptibility, fruit quality, and yield. Evaluation methods for these\ntraits include visual scoring, manual methodologies, and computer vision, with\nthe latter being the most scalable approach. Most of the existing computer\nvision approaches for processing cluster images often rely on conventional\nsegmentation or machine learning with extensive training and limited\ngeneralization. The Segment Anything Model (SAM), a novel foundation model\ntrained on a massive image dataset, enables automated object segmentation\nwithout additional training. This study demonstrates out-of-the-box SAM's high\naccuracy in identifying individual berries in 2D cluster images. Using this\nmodel, we managed to segment approximately 3,500 cluster images, generating\nover 150,000 berry masks, each linked with spatial coordinates within their\nclusters. The correlation between human-identified berries and SAM predictions\nwas very strong (Pearson r2=0.96). Although the visible berry count in images\ntypically underestimates the actual cluster berry count due to visibility\nissues, we demonstrated that this discrepancy could be adjusted using a linear\nregression model (adjusted R2=0.87). We emphasized the critical importance of\nthe angle at which the cluster is imaged, noting its substantial effect on\nberry counts and architecture. We proposed different approaches in which berry\nlocation information facilitated the calculation of complex features related to\ncluster architecture and compactness. Finally, we discussed SAM's potential\nintegration into currently available pipelines for image generation and\nprocessing in vineyard conditions.\n", "date": "2024-03-19", "categories": []}
{"id": "2403.07137", "title": "Exploring Cluster Analysis in Nelore Cattle Visual Score Attribution", "abstract": "  Assessing the biotype of cattle through human visual inspection is a very\ncommon and important practice in precision cattle breeding. This paper presents\nthe results of a correlation analysis between scores produced by humans for\nNelore cattle and a variety of measurements that can be derived from images or\nother instruments. It also presents a study using the k-means algorithm to\ngenerate new ways of clustering a batch of cattle using the measurements that\nmost correlate with the animal's body weight and visual scores.\n", "date": "2024-03-11", "categories": []}
{"id": "2403.03390", "title": "Performance Evaluation of Semi-supervised Learning Frameworks for\n  Multi-Class Weed Detection", "abstract": "  Effective weed control plays a crucial role in optimizing crop yield and\nenhancing agricultural product quality. However, the reliance on herbicide\napplication not only poses a critical threat to the environment but also\npromotes the emergence of resistant weeds. Fortunately, recent advances in\nprecision weed management enabled by ML and DL provide a sustainable\nalternative. Despite great progress, existing algorithms are mainly developed\nbased on supervised learning approaches, which typically demand large-scale\ndatasets with manual-labeled annotations, which is time-consuming and\nlabor-intensive. As such, label-efficient learning methods, especially\nsemi-supervised learning, have gained increased attention in the broader domain\nof computer vision and have demonstrated promising performance. These methods\naim to utilize a small number of labeled data samples along with a great number\nof unlabeled samples to develop high-performing models comparable to the\nsupervised learning counterpart trained on a large amount of labeled data\nsamples. In this study, we assess the effectiveness of a semi-supervised\nlearning framework for multi-class weed detection, employing two well-known\nobject detection frameworks, namely FCOS and Faster-RCNN. Specifically, we\nevaluate a generalized student-teacher framework with an improved pseudo-label\ngeneration module to produce reliable pseudo-labels for the unlabeled data. To\nenhance generalization, an ensemble student network is employed to facilitate\nthe training process. Experimental results show that the proposed approach is\nable to achieve approximately 76\\% and 96\\% detection accuracy as the\nsupervised methods with only 10\\% of labeled data in CottenWeedDet3 and\nCottonWeedDet12, respectively. We offer access to the source code, contributing\na valuable resource for ongoing semi-supervised learning research in weed\ndetection and beyond.\n", "date": "2024-03-06", "categories": []}
{"id": "2402.07059", "title": "Domain Adaptable Fine-Tune Distillation Framework For Advancing Farm\n  Surveillance", "abstract": "  In this study, we propose an automated framework for camel farm monitoring,\nintroducing two key contributions: the Unified Auto-Annotation framework and\nthe Fine-Tune Distillation framework. The Unified Auto-Annotation approach\ncombines two models, GroundingDINO (GD), and Segment-Anything-Model (SAM), to\nautomatically annotate raw datasets extracted from surveillance videos.\nBuilding upon this foundation, the Fine-Tune Distillation framework conducts\nfine-tuning of student models using the auto-annotated dataset. This process\ninvolves transferring knowledge from a large teacher model to a student model,\nresembling a variant of Knowledge Distillation. The Fine-Tune Distillation\nframework aims to be adaptable to specific use cases, enabling the transfer of\nknowledge from the large models to the small models, making it suitable for\ndomain-specific applications. By leveraging our raw dataset collected from\nAl-Marmoom Camel Farm in Dubai, UAE, and a pre-trained teacher model,\nGroundingDINO, the Fine-Tune Distillation framework produces a lightweight\ndeployable model, YOLOv8. This framework demonstrates high performance and\ncomputational efficiency, facilitating efficient real-time object detection.\nOur code is available at\n\\href{https://github.com/Razaimam45/Fine-Tune-Distillation}{https://github.com/Razaimam45/Fine-Tune-Distillation}\n", "date": "2024-02-10", "categories": []}
{"id": "2402.02121", "title": "Enhancing crop classification accuracy by synthetic SAR-Optical data\n  generation using deep learning", "abstract": "  Crop classification using remote sensing data has emerged as a prominent\nresearch area in recent decades. Studies have demonstrated that fusing SAR and\noptical images can significantly enhance the accuracy of classification.\nHowever, a major challenge in this field is the limited availability of\ntraining data, which adversely affects the performance of classifiers. In\nagricultural regions, the dominant crops typically consist of one or two\nspecific types, while other crops are scarce. Consequently, when collecting\ntraining samples to create a map of agricultural products, there is an\nabundance of samples from the dominant crops, forming the majority classes.\nConversely, samples from other crops are scarce, representing the minority\nclasses. Addressing this issue requires overcoming several challenges and\nweaknesses associated with traditional data generation methods. These methods\nhave been employed to tackle the imbalanced nature of the training data.\nNevertheless, they still face limitations in effectively handling the minority\nclasses. Overall, the issue of inadequate training data, particularly for\nminority classes, remains a hurdle that traditional methods struggle to\novercome. In this research, We explore the effectiveness of conditional tabular\ngenerative adversarial network (CTGAN) as a synthetic data generation method\nbased on a deep learning network, in addressing the challenge of limited\ntraining data for minority classes in crop classification using the fusion of\nSAR-optical data. Our findings demonstrate that the proposed method generates\nsynthetic data with higher quality that can significantly increase the number\nof samples for minority classes leading to better performance of crop\nclassifiers.\n", "date": "2024-02-03", "categories": []}
{"id": "2311.03076", "title": "SugarViT -- Multi-objective Regression of UAV Images with Vision\n  Transformers and Deep Label Distribution Learning Demonstrated on Disease\n  Severity Prediction in Sugar Beet", "abstract": "  Remote sensing and artificial intelligence are pivotal technologies of\nprecision agriculture nowadays. The efficient retrieval of large-scale field\nimagery combined with machine learning techniques shows success in various\ntasks like phenotyping, weeding, cropping, and disease control. This work will\nintroduce a machine learning framework for automatized large-scale\nplant-specific trait annotation for the use case disease severity scoring for\nCercospora Leaf Spot (CLS) in sugar beet. With concepts of Deep Label\nDistribution Learning (DLDL), special loss functions, and a tailored model\narchitecture, we develop an efficient Vision Transformer based model for\ndisease severity scoring called SugarViT. One novelty in this work is the\ncombination of remote sensing data with environmental parameters of the\nexperimental sites for disease severity prediction. Although the model is\nevaluated on this special use case, it is held as generic as possible to also\nbe applicable to various image-based classification and regression tasks. With\nour framework, it is even possible to learn models on multi-objective problems\nas we show by a pretraining on environmental metadata.\n", "date": "2023-11-06", "categories": []}
{"id": "2401.15785", "title": "Real-time object detection and robotic manipulation for agriculture\n  using a YOLO-based learning approach", "abstract": "  The optimisation of crop harvesting processes for commonly cultivated crops\nis of great importance in the aim of agricultural industrialisation. Nowadays,\nthe utilisation of machine vision has enabled the automated identification of\ncrops, leading to the enhancement of harvesting efficiency, but challenges\nstill exist. This study presents a new framework that combines two separate\narchitectures of convolutional neural networks (CNNs) in order to\nsimultaneously accomplish the tasks of crop detection and harvesting (robotic\nmanipulation) inside a simulated environment. Crop images in the simulated\nenvironment are subjected to random rotations, cropping, brightness, and\ncontrast adjustments to create augmented images for dataset generation. The you\nonly look once algorithmic framework is employed with traditional rectangular\nbounding boxes for crop localization. The proposed method subsequently utilises\nthe acquired image data via a visual geometry group model in order to reveal\nthe grasping positions for the robotic manipulation.\n", "date": "2024-01-28", "categories": []}
{"id": "2308.15690", "title": "CongNaMul: A Dataset for Advanced Image Processing of Soybean Sprouts", "abstract": "  We present 'CongNaMul', a comprehensive dataset designed for various tasks in\nsoybean sprouts image analysis. The CongNaMul dataset is curated to facilitate\ntasks such as image classification, semantic segmentation, decomposition, and\nmeasurement of length and weight. The classification task provides four classes\nto determine the quality of soybean sprouts: normal, broken, spotted, and\nbroken and spotted, for the development of AI-aided automatic quality\ninspection technology. For semantic segmentation, images with varying\ncomplexity, from single sprout images to images with multiple sprouts, along\nwith human-labelled mask images, are included. The label has 4 different\nclasses: background, head, body, tail. The dataset also provides images and\nmasks for the image decomposition task, including two separate sprout images\nand their combined form. Lastly, 5 physical features of sprouts (head length,\nbody length, body thickness, tail length, weight) are provided for image-based\nmeasurement tasks. This dataset is expected to be a valuable resource for a\nwide range of research and applications in the advanced analysis of images of\nsoybean sprouts. Also, we hope that this dataset can assist researchers\nstudying classification, semantic segmentation, decomposition, and physical\nfeature measurement in other industrial fields, in evaluating their models. The\ndataset is available at the authors' repository. (https://bhban.kr/data)\n", "date": "2023-08-30", "categories": []}
{"id": "2401.15055", "title": "Deep learning-based approach for tomato classification in complex scenes", "abstract": "  Tracking ripening tomatoes is time consuming and labor intensive. Artificial\nintelligence technologies combined with those of computer vision can help users\noptimize the process of monitoring the ripening status of plants. To this end,\nwe have proposed a tomato ripening monitoring approach based on deep learning\nin complex scenes. The objective is to detect mature tomatoes and harvest them\nin a timely manner. The proposed approach is declined in two parts. Firstly,\nthe images of the scene are transmitted to the pre-processing layer. This\nprocess allows the detection of areas of interest (area of the image containing\ntomatoes). Then, these images are used as input to the maturity detection\nlayer. This layer, based on a deep neural network learning algorithm,\nclassifies the tomato thumbnails provided to it in one of the following five\ncategories: green, brittle, pink, pale red, mature red. The experiments are\nbased on images collected from the internet gathered through searches using\ntomato state across diverse languages including English, German, French, and\nSpanish. The experimental results of the maturity detection layer on a dataset\ncomposed of images of tomatoes taken under the extreme conditions, gave a good\nclassification rate.\n", "date": "2024-01-26", "categories": []}
{"id": "2401.10379", "title": "Agricultural Object Detection with You Look Only Once (YOLO) Algorithm:\n  A Bibliometric and Systematic Literature Review", "abstract": "  Vision is a major component in several digital technologies and tools used in\nagriculture. The object detector, You Look Only Once (YOLO), has gained\npopularity in agriculture in a relatively short span due to its\nstate-of-the-art performance. YOLO offers real-time detection with good\naccuracy and is implemented in various agricultural tasks, including\nmonitoring, surveillance, sensing, automation, and robotics. The research and\napplication of YOLO in agriculture are accelerating rapidly but are fragmented\nand multidisciplinary. Moreover, the performance characteristics (i.e.,\naccuracy, speed, computation) of the object detector influence the rate of\ntechnology implementation and adoption in agriculture. Thus, the study aims to\ncollect extensive literature to document and critically evaluate the advances\nand application of YOLO for agricultural object recognition. First, we\nconducted a bibliometric review of 257 articles to understand the scholarly\nlandscape of YOLO in agricultural domain. Secondly, we conducted a systematic\nreview of 30 articles to identify current knowledge, gaps, and modifications in\nYOLO for specific agricultural tasks. The study critically assesses and\nsummarizes the information on YOLO's end-to-end learning approach, including\ndata acquisition, processing, network modification, integration, and\ndeployment. We also discussed task-specific YOLO algorithm modification and\nintegration to meet the agricultural object or environment-specific challenges.\nIn general, YOLO-integrated digital tools and technologies show the potential\nfor real-time, automated monitoring, surveillance, and object handling to\nreduce labor, production cost, and environmental impact while maximizing\nresource efficiency. The study provides detailed documentation and\nsignificantly advances the existing knowledge on applying YOLO in agriculture,\nwhich can greatly benefit the scientific community.\n", "date": "2024-01-18", "categories": []}
{"id": "2401.08720", "title": "Unsupervised Pre-Training for 3D Leaf Instance Segmentation", "abstract": "  Crops for food, feed, fiber, and fuel are key natural resources for our\nsociety. Monitoring plants and measuring their traits is an important task in\nagriculture often referred to as plant phenotyping. Traditionally, this task is\ndone manually, which is time- and labor-intensive. Robots can automate\nphenotyping providing reproducible and high-frequency measurements. Today's\nperception systems use deep learning to interpret these measurements, but\nrequire a substantial amount of annotated data to work well. Obtaining such\nlabels is challenging as it often requires background knowledge on the side of\nthe labelers. This paper addresses the problem of reducing the labeling effort\nrequired to perform leaf instance segmentation on 3D point clouds, which is a\nfirst step toward phenotyping in 3D. Separating all leaves allows us to count\nthem and compute relevant traits as their areas, lengths, and widths. We\npropose a novel self-supervised task-specific pre-training approach to\ninitialize the backbone of a network for leaf instance segmentation. We also\nintroduce a novel automatic postprocessing that considers the difficulty of\ncorrectly segmenting the points close to the stem, where all the leaves petiole\noverlap. The experiments presented in this paper suggest that our approach\nboosts the performance over all the investigated scenarios. We also evaluate\nthe embeddings to assess the quality of the fully unsupervised approach and see\na higher performance of our domain-specific postprocessing.\n", "date": "2024-01-16", "categories": []}
{"id": "2401.07014", "title": "Weak Labeling for Cropland Mapping in Africa", "abstract": "  Cropland mapping can play a vital role in addressing environmental,\nagricultural, and food security challenges. However, in the context of Africa,\npractical applications are often hindered by the limited availability of\nhigh-resolution cropland maps. Such maps typically require extensive human\nlabeling, thereby creating a scalability bottleneck. To address this, we\npropose an approach that utilizes unsupervised object clustering to refine\nexisting weak labels, such as those obtained from global cropland maps. The\nrefined labels, in conjunction with sparse human annotations, serve as training\ndata for a semantic segmentation network designed to identify cropland areas.\nWe conduct experiments to demonstrate the benefits of the improved weak labels\ngenerated by our method. In a scenario where we train our model with only 33\nhuman-annotated labels, the F_1 score for the cropland category increases from\n0.53 to 0.84 when we add the mined negative labels.\n", "date": "2024-01-13", "categories": []}
{"id": "2401.05768", "title": "Evaluating Data Augmentation Techniques for Coffee Leaf Disease\n  Classification", "abstract": "  The detection and classification of diseases in Robusta coffee leaves are\nessential to ensure that plants are healthy and the crop yield is kept high.\nHowever, this job requires extensive botanical knowledge and much wasted time.\nTherefore, this task and others similar to it have been extensively researched\nsubjects in image classification. Regarding leaf disease classification, most\napproaches have used the more popular PlantVillage dataset while completely\ndisregarding other datasets, like the Robusta Coffee Leaf (RoCoLe) dataset. As\nthe RoCoLe dataset is imbalanced and does not have many samples, fine-tuning of\npre-trained models and multiple augmentation techniques need to be used. The\ncurrent paper uses the RoCoLe dataset and approaches based on deep learning for\nclassifying coffee leaf diseases from images, incorporating the pix2pix model\nfor segmentation and cycle-generative adversarial network (CycleGAN) for\naugmentation. Our study demonstrates the effectiveness of Transformer-based\nmodels, online augmentations, and CycleGAN augmentation in improving leaf\ndisease classification. While synthetic data has limitations, it complements\nreal data, enhancing model performance. These findings contribute to developing\nrobust techniques for plant disease detection and classification.\n", "date": "2024-01-11", "categories": []}
{"id": "2401.04750", "title": "DedustNet: A Frequency-dominated Swin Transformer-based Wavelet Network\n  for Agricultural Dust Removal", "abstract": "  While dust significantly affects the environmental perception of automated\nagricultural machines, the existing deep learning-based methods for dust\nremoval require further research and improvement in this area to improve the\nperformance and reliability of automated agricultural machines in agriculture.\nWe propose an end-to-end trainable learning network (DedustNet) to solve the\nreal-world agricultural dust removal task. To our knowledge, DedustNet is the\nfirst time Swin Transformer-based units have been used in wavelet networks for\nagricultural image dusting. Specifically, we present the frequency-dominated\nblock (DWTFormer block and IDWTFormer block) by adding a spatial features\naggregation scheme (SFAS) to the Swin Transformer and combining it with the\nwavelet transform, the DWTFormer block and IDWTFormer block, alleviating the\nlimitation of the global receptive field of Swin Transformer when dealing with\ncomplex dusty backgrounds. Furthermore, We propose a cross-level information\nfusion module to fuse different levels of features and effectively capture\nglobal and long-range feature relationships. In addition, we present a dilated\nconvolution module to capture contextual information guided by wavelet\ntransform at multiple scales, which combines the advantages of wavelet\ntransform and dilated convolution. Our algorithm leverages deep learning\ntechniques to effectively remove dust from images while preserving the original\nstructural and textural features. Compared to existing state-of-the-art\nmethods, DedustNet achieves superior performance and more reliable results in\nagricultural image dedusting, providing strong support for the application of\nagricultural machinery in dusty environments. Additionally, the impressive\nperformance on real-world hazy datasets and application tests highlights\nDedustNet superior generalization ability and computer vision-related\napplication performance.\n", "date": "2024-01-09", "categories": []}
{"id": "2312.16331", "title": "Early and Accurate Detection of Tomato Leaf Diseases Using TomFormer", "abstract": "  Tomato leaf diseases pose a significant challenge for tomato farmers,\nresulting in substantial reductions in crop productivity. The timely and\nprecise identification of tomato leaf diseases is crucial for successfully\nimplementing disease management strategies. This paper introduces a\ntransformer-based model called TomFormer for the purpose of tomato leaf disease\ndetection. The paper's primary contributions include the following: Firstly, we\npresent a novel approach for detecting tomato leaf diseases by employing a\nfusion model that combines a visual transformer and a convolutional neural\nnetwork. Secondly, we aim to apply our proposed methodology to the Hello\nStretch robot to achieve real-time diagnosis of tomato leaf diseases. Thirdly,\nwe assessed our method by comparing it to models like YOLOS, DETR, ViT, and\nSwin, demonstrating its ability to achieve state-of-the-art outcomes. For the\npurpose of the experiment, we used three datasets of tomato leaf diseases,\nnamely KUTomaDATA, PlantDoc, and PlanVillage, where KUTomaDATA is being\ncollected from a greenhouse in Abu Dhabi, UAE. Finally, we present a\ncomprehensive analysis of the performance of our model and thoroughly discuss\nthe limitations inherent in our approach. TomFormer performed well on the\nKUTomaDATA, PlantDoc, and PlantVillage datasets, with mean average accuracy\n(mAP) scores of 87%, 81%, and 83%, respectively. The comparative results in\nterms of mAP demonstrate that our method exhibits robustness, accuracy,\nefficiency, and scalability. Furthermore, it can be readily adapted to new\ndatasets. We are confident that our work holds the potential to significantly\ninfluence the tomato industry by effectively mitigating crop losses and\nenhancing crop yields.\n", "date": "2023-12-26", "categories": []}
{"id": "2312.16724", "title": "A pipeline for multiple orange detection and tracking with 3-D fruit\n  relocalization and neural-net based yield regression in commercial citrus\n  orchards", "abstract": "  Traditionally, sweet orange crop forecasting has involved manually counting\nfruits from numerous trees, which is a labor-intensive process. Automatic\nsystems for fruit counting, based on proximal imaging, computer vision, and\nmachine learning, have been considered a promising alternative or complement to\nmanual counting. These systems require data association components that prevent\nmultiple counting of the same fruit observed in different images. However,\nthere is a lack of work evaluating the accuracy of multiple fruit counting,\nespecially considering (i) occluded and re-entering green fruits on leafy\ntrees, and (ii) counting ground-truth data measured in the crop field. We\npropose a non-invasive alternative that utilizes fruit counting from videos,\nimplemented as a pipeline. Firstly, we employ CNNs for the detection of visible\nfruits. Inter-frame association techniques are then applied to track the fruits\nacross frames. To handle occluded and re-appeared fruit, we introduce a\nrelocalization component that employs 3-D estimation of fruit locations.\nFinally, a neural network regressor is utilized to estimate the total number of\nfruit, integrating image-based fruit counting with other tree data such as crop\nvariety and tree size. The results demonstrate that the performance of our\napproach is closely tied to the quality of the field-collected videos. By\nensuring that at least 30% of the fruit is accurately detected, tracked, and\ncounted, our yield regressor achieves an impressive coefficient of\ndetermination of 0.85. To the best of our knowledge, this study represents one\nof the few endeavors in fruit estimation that incorporates manual fruit\ncounting as a reference point for evaluation. We also introduce annotated\ndatasets for multiple orange tracking (MOrangeT) and detection (OranDet),\npublicly available to foster the development of novel methods for image-based\nfruit counting.\n", "date": "2023-12-27", "categories": []}
{"id": "2312.10922", "title": "NTrack: A Multiple-Object Tracker and Dataset for Infield Cotton Boll\n  Counting", "abstract": "  In agriculture, automating the accurate tracking of fruits, vegetables, and\nfiber is a very tough problem. The issue becomes extremely challenging in\ndynamic field environments. Yet, this information is critical for making\nday-to-day agricultural decisions, assisting breeding programs, and much more.\nTo tackle this dilemma, we introduce NTrack, a novel multiple object tracking\nframework based on the linear relationship between the locations of neighboring\ntracks. NTrack computes dense optical flow and utilizes particle filtering to\nguide each tracker. Correspondences between detections and tracks are found\nthrough data association via direct observations and indirect cues, which are\nthen combined to obtain an updated observation. Our modular multiple object\ntracking system is independent of the underlying detection method, thus\nallowing for the interchangeable use of any off-the-shelf object detector. We\nshow the efficacy of our approach on the task of tracking and counting infield\ncotton bolls. Experimental results show that our system exceeds contemporary\ntracking and cotton boll-based counting methods by a large margin. Furthermore,\nwe publicly release the first annotated cotton boll video dataset to the\nresearch community.\n", "date": "2023-12-18", "categories": []}
{"id": "2312.08408", "title": "Explainable AI in Grassland Monitoring: Enhancing Model Performance and\n  Domain Adaptability", "abstract": "  Grasslands are known for their high biodiversity and ability to provide\nmultiple ecosystem services. Challenges in automating the identification of\nindicator plants are key obstacles to large-scale grassland monitoring. These\nchallenges stem from the scarcity of extensive datasets, the distributional\nshifts between generic and grassland-specific datasets, and the inherent\nopacity of deep learning models. This paper delves into the latter two\nchallenges, with a specific focus on transfer learning and eXplainable\nArtificial Intelligence (XAI) approaches to grassland monitoring, highlighting\nthe novelty of XAI in this domain. We analyze various transfer learning methods\nto bridge the distributional gaps between generic and grassland-specific\ndatasets. Additionally, we showcase how explainable AI techniques can unveil\nthe model's domain adaptation capabilities, employing quantitative assessments\nto evaluate the model's proficiency in accurately centering relevant input\nfeatures around the object of interest. This research contributes valuable\ninsights for enhancing model performance through transfer learning and\nmeasuring domain adaptability with explainable AI, showing significant promise\nfor broader applications within the agricultural community.\n", "date": "2023-12-13", "categories": []}
{"id": "2312.08805", "title": "Zoom in on the Plant: Fine-grained Analysis of Leaf, Stem and Vein\n  Instances", "abstract": "  Robot perception is far from what humans are capable of. Humans do not only\nhave a complex semantic scene understanding but also extract fine-grained\nintra-object properties for the salient ones. When humans look at plants, they\nnaturally perceive the plant architecture with its individual leaves and\nbranching system. In this work, we want to advance the granularity in plant\nunderstanding for agricultural precision robots. We develop a model to extract\nfine-grained phenotypic information, such as leaf-, stem-, and vein instances.\nThe underlying dataset RumexLeaves is made publicly available and is the first\nof its kind with keypoint-guided polyline annotations leading along the line\nfrom the lowest stem point along the leaf basal to the leaf apex. Furthermore,\nwe introduce an adapted metric POKS complying with the concept of\nkeypoint-guided polylines. In our experimental evaluation, we provide baseline\nresults for our newly introduced dataset while showcasing the benefits of POKS\nover OKS.\n", "date": "2023-12-14", "categories": []}
{"id": "2312.07905", "title": "Plant Disease Recognition Datasets in the Age of Deep Learning:\n  Challenges and Opportunities", "abstract": "  Plant disease recognition has witnessed a significant improvement with deep\nlearning in recent years. Although plant disease datasets are essential and\nmany relevant datasets are public available, two fundamental questions exist.\nFirst, how to differentiate datasets and further choose suitable public\ndatasets for specific applications? Second, what kinds of characteristics of\ndatasets are desired to achieve promising performance in real-world\napplications? To address the questions, this study explicitly propose an\ninformative taxonomy to describe potential plant disease datasets. We further\nprovide several directions for future, such as creating challenge-oriented\ndatasets and the ultimate objective deploying deep learning in real-world\napplications with satisfactory performance. In addition, existing related\npublic RGB image datasets are summarized. We believe that this study will\ncontributing making better datasets and that this study will contribute beyond\nplant disease recognition such as plant species recognition. To facilitate the\ncommunity, our project is public https://github.com/xml94/PPDRD with the\ninformation of relevant public datasets.\n", "date": "2023-12-13", "categories": []}
{"id": "2212.01506", "title": "Autonomous Apple Fruitlet Sizing and Growth Rate Tracking using Computer\n  Vision", "abstract": "  In this paper, we present a computer vision-based approach to measure the\nsizes and growth rates of apple fruitlets. Measuring the growth rates of apple\nfruitlets is important because it allows apple growers to determine when to\napply chemical thinners to their crops in order to optimize yield. The current\npractice of obtaining growth rates involves using calipers to record sizes of\nfruitlets across multiple days. Due to the number of fruitlets needed to be\nsized, this method is laborious, time-consuming, and prone to human error. With\nimages collected by a hand-held stereo camera, our system, segments, clusters,\nand fits ellipses to fruitlets to measure their diameters. The growth rates are\nthen calculated by temporally associating clustered fruitlets across days. We\nprovide quantitative results on data collected in an apple orchard, and\ndemonstrate that our system is able to predict abscise rates within 3.5% of the\ncurrent method with a 6 times improvement in speed, while requiring\nsignificantly less manual effort. Moreover, we provide results on images\ncaptured by a robotic system in the field, and discuss the next steps required\nto make the process fully autonomous.\n", "date": "2022-12-03", "categories": []}
{"id": "2211.02760", "title": "Development and evaluation of automated localisation and reconstruction\n  of all fruits on tomato plants in a greenhouse based on multi-view perception\n  and 3D multi-object tracking", "abstract": "  The ability to accurately represent and localise relevant objects is\nessential for robots to carry out tasks effectively. Traditional approaches,\nwhere robots simply capture an image, process that image to take an action, and\nthen forget the information, have proven to struggle in the presence of\nocclusions. Methods using multi-view perception, which have the potential to\naddress some of these problems, require a world model that guides the\ncollection, integration and extraction of information from multiple viewpoints.\nFurthermore, constructing a generic representation that can be applied in\nvarious environments and tasks is a difficult challenge. In this paper, a novel\napproach for building generic representations in occluded agro-food\nenvironments using multi-view perception and 3D multi-object tracking is\nintroduced. The method is based on a detection algorithm that generates partial\npoint clouds for each detected object, followed by a 3D multi-object tracking\nalgorithm that updates the representation over time. The accuracy of the\nrepresentation was evaluated in a real-world environment, where successful\nrepresentation and localisation of tomatoes in tomato plants were achieved,\ndespite high levels of occlusion, with the total count of tomatoes estimated\nwith a maximum error of 5.08% and the tomatoes tracked with an accuracy up to\n71.47%. Novel tracking metrics were introduced, demonstrating that valuable\ninsight into the errors in localising and representing the fruits can be\nprovided by their use. This approach presents a novel solution for building\nrepresentations in occluded agro-food environments, demonstrating potential to\nenable robots to perform tasks effectively in these challenging environments.\n", "date": "2022-11-04", "categories": []}
{"id": "2311.12993", "title": "AI for Agriculture: the Comparison of Semantic Segmentation Methods for\n  Crop Mapping with Sentinel-2 Imagery", "abstract": "  Crop mapping is one of the most common tasks in artificial intelligence for\nagriculture due to higher food demands from a growing population and increased\nawareness of climate change. In case of vineyards, the texture is very\nimportant for crop segmentation: with higher resolution satellite imagery the\ntexture is easily detected by majority of state-of-the-art algorithms. However,\nthis task becomes increasingly more difficult as the resolution of satellite\nimagery decreases and the information about the texture becomes unavailable. In\nthis paper we aim to explore the main machine learning methods that can be used\nwith freely available satellite imagery and discuss how and when they can be\napplied for vineyard segmentation problem. We assess the effectiveness of\nvarious widely-used machine learning techniques and offer guidance on selecting\nthe most suitable model for specific scenarios.\n", "date": "2023-11-21", "categories": []}
{"id": "2310.10861", "title": "SoybeanNet: Transformer-Based Convolutional Neural Network for Soybean\n  Pod Counting from Unmanned Aerial Vehicle (UAV) Images", "abstract": "  Soybeans are a critical source of food, protein and oil, and thus have\nreceived extensive research aimed at enhancing their yield, refining\ncultivation practices, and advancing soybean breeding techniques. Within this\ncontext, soybean pod counting plays an essential role in understanding and\noptimizing production. Despite recent advancements, the development of a robust\npod-counting algorithm capable of performing effectively in real-field\nconditions remains a significant challenge This paper presents a pioneering\nwork of accurate soybean pod counting utilizing unmanned aerial vehicle (UAV)\nimages captured from actual soybean fields in Michigan, USA. Specifically, this\npaper presents SoybeanNet, a novel point-based counting network that harnesses\npowerful transformer backbones for simultaneous soybean pod counting and\nlocalization with high accuracy. In addition, a new dataset of UAV-acquired\nimages for soybean pod counting was created and open-sourced, consisting of 113\ndrone images with more than 260k manually annotated soybean pods captured under\nnatural lighting conditions. Through comprehensive evaluations, SoybeanNet\ndemonstrated superior performance over five state-of-the-art approaches when\ntested on the collected images. Remarkably, SoybeanNet achieved a counting\naccuracy of $84.51\\%$ when tested on the testing dataset, attesting to its\nefficacy in real-world scenarios. The publication also provides both the source\ncode (\\url{https://github.com/JiajiaLi04/Soybean-Pod-Counting-from-UAV-Images})\nand the labeled soybean dataset\n(\\url{https://www.kaggle.com/datasets/jiajiali/uav-based-soybean-pod-images}),\noffering a valuable resource for future research endeavors in soybean pod\ncounting and related fields.\n", "date": "2023-10-16", "categories": []}
{"id": "2311.11901", "title": "Identifying the Defective: Detecting Damaged Grains for Cereal\n  Appearance Inspection", "abstract": "  Cereal grain plays a crucial role in the human diet as a major source of\nessential nutrients. Grain Appearance Inspection (GAI) serves as an essential\nprocess to determine grain quality and facilitate grain circulation and\nprocessing. However, GAI is routinely performed manually by inspectors with\ncumbersome procedures, which poses a significant bottleneck in smart\nagriculture.\n  In this paper, we endeavor to develop an automated GAI system:AI4GrainInsp.\nBy analyzing the distinctive characteristics of grain kernels, we formulate GAI\nas a ubiquitous problem: Anomaly Detection (AD), in which healthy and edible\nkernels are considered normal samples while damaged grains or unknown objects\nare regarded as anomalies. We further propose an AD model, called AD-GAI, which\nis trained using only normal samples yet can identify anomalies during\ninference. Moreover, we customize a prototype device for data acquisition and\ncreate a large-scale dataset including 220K high-quality images of wheat and\nmaize kernels. Through extensive experiments, AD-GAI achieves considerable\nperformance in comparison with advanced AD methods, and AI4GrainInsp has highly\nconsistent performance compared to human experts and excels at inspection\nefficiency over 20x speedup. The dataset, code and models will be released at\nhttps://github.com/hellodfan/AI4GrainInsp.\n", "date": "2023-11-20", "categories": []}
{"id": "2311.08284", "title": "Level Set KSVD", "abstract": "  We present a new algorithm for image segmentation - Level-set KSVD. Level-set\nKSVD merges the methods of sparse dictionary learning for feature extraction\nand variational level-set method for image segmentation. Specifically, we use a\ngeneralization of the Chan-Vese functional with features learned by KSVD. The\nmotivation for this model is agriculture based. Aerial images are taken in\norder to detect the spread of fungi in various crops. Our model is tested on\nsuch images of cotton fields. The results are compared to other methods.\n", "date": "2023-11-14", "categories": []}
{"id": "2311.04699", "title": "3D Pose Estimation of Tomato Peduncle Nodes using Deep Keypoint\n  Detection and Point Cloud", "abstract": "  Greenhouse production of fruits and vegetables in developed countries is\nchallenged by labor 12 scarcity and high labor costs. Robots offer a good\nsolution for sustainable and cost-effective 13 production. Acquiring accurate\nspatial information about relevant plant parts is vital for 14 successful robot\noperation. Robot perception in greenhouses is challenging due to variations in\n15 plant appearance, viewpoints, and illumination. This paper proposes a\nkeypoint-detection-based 16 method using data from an RGB-D camera to estimate\nthe 3D pose of peduncle nodes, which 17 provides essential information to\nharvest the tomato bunches. 18 19 Specifically, this paper proposes a method\nthat detects four anatomical landmarks in the color 20 image and then\nintegrates 3D point-cloud information to determine the 3D pose. A 21\ncomprehensive evaluation was conducted in a commercial greenhouse to gain\ninsight into the 22 performance of different parts of the method. The results\nshowed: (1) high accuracy in object 23 detection, achieving an Average\nPrecision (AP) of AP@0.5=0.96; (2) an average Percentage of 24 Detected Joints\n(PDJ) of the keypoints of PhDJ@0.2=94.31%; and (3) 3D pose estimation 25\naccuracy with mean absolute errors (MAE) of 11.38o and 9.93o for the relative\nupper and lower 26 angles between the peduncle and main stem, respectively.\nFurthermore, the capability to handle 27 variations in viewpoint was\ninvestigated, demonstrating the method was robust to view changes. 28 However,\ncanonical and higher views resulted in slightly higher performance compared to\nother 29 views. Although tomato was selected as a use case, the proposed method\nis also applicable to 30 other greenhouse crops like pepper.\n", "date": "2023-11-08", "categories": []}
{"id": "2311.00429", "title": "Crop Disease Classification using Support Vector Machines with Green\n  Chromatic Coordinate (GCC) and Attention based feature extraction for IoT\n  based Smart Agricultural Applications", "abstract": "  Crops hold paramount significance as they serve as the primary provider of\nenergy, nutrition, and medicinal benefits for the human population. Plant\ndiseases, however, can negatively affect leaves during agricultural\ncultivation, resulting in significant losses in crop output and economic value.\nTherefore, it is crucial for farmers to identify crop diseases. However, this\nmethod frequently necessitates hard work, a lot of planning, and in-depth\nfamiliarity with plant pathogens. Given these numerous obstacles, it is\nessential to provide solutions that can easily interface with mobile and IoT\ndevices so that our farmers can guarantee the best possible crop development.\nVarious machine learning (ML) as well as deep learning (DL) algorithms have\nbeen created & studied for the identification of plant disease detection,\nyielding substantial and promising results. This article presents a novel\nclassification method that builds on prior work by utilising attention-based\nfeature extraction, RGB channel-based chromatic analysis, Support Vector\nMachines (SVM) for improved performance, and the ability to integrate with\nmobile applications and IoT devices after quantization of information. Several\ndisease classification algorithms were compared with the suggested model, and\nit was discovered that, in terms of accuracy, Vision Transformer-based feature\nextraction and additional Green Chromatic Coordinate feature with SVM\nclassification achieved an accuracy of (GCCViT-SVM) - 99.69%, whereas after\nquantization for IoT device integration achieved an accuracy of - 97.41% while\nalmost reducing 4x in size. Our findings have profound implications because\nthey have the potential to transform how farmers identify crop illnesses with\nprecise and fast information, thereby preserving agricultural output and\nensuring food security.\n", "date": "2023-11-01", "categories": []}
{"id": "2310.19830", "title": "GalliformeSpectra: A Hen Breed Dataset", "abstract": "  This article presents a comprehensive dataset featuring ten distinct hen\nbreeds, sourced from various regions, capturing the unique characteristics and\ntraits of each breed. The dataset encompasses Bielefeld, Blackorpington,\nBrahma, Buckeye, Fayoumi, Leghorn, Newhampshire, Plymouthrock, Sussex, and\nTurken breeds, offering a diverse representation of poultry commonly bred\nworldwide. A total of 1010 original JPG images were meticulously collected,\nshowcasing the physical attributes, feather patterns, and distinctive features\nof each hen breed. These images were subsequently standardized, resized, and\nconverted to PNG format for consistency within the dataset. The compilation,\nalthough unevenly distributed across the breeds, provides a rich resource,\nserving as a foundation for research and applications in poultry science,\ngenetics, and agricultural studies. This dataset holds significant potential to\ncontribute to various fields by enabling the exploration and analysis of unique\ncharacteristics and genetic traits across different hen breeds, thereby\nsupporting advancements in poultry breeding, farming, and genetic research.\n", "date": "2023-10-28", "categories": []}
{"id": "2310.18921", "title": "QWID: Quantized Weed Identification Deep neural network", "abstract": "  In this paper, we present an efficient solution for weed classification in\nagriculture. We focus on optimizing model performance at inference while\nrespecting the constraints of the agricultural domain. We propose a Quantized\nDeep Neural Network model that classifies a dataset of 9 weed classes using\n8-bit integer (int8) quantization, a departure from standard 32-bit floating\npoint (fp32) models. Recognizing the hardware resource limitations in\nagriculture, our model balances model size, inference time, and accuracy,\naligning with practical requirements. We evaluate the approach on ResNet-50 and\nInceptionV3 architectures, comparing their performance against their int8\nquantized versions. Transfer learning and fine-tuning are applied using the\nDeepWeeds dataset. The results show staggering model size and inference time\nreductions while maintaining accuracy in real-world production scenarios like\nDesktop, Mobile and Raspberry Pi. Our work sheds light on a promising direction\nfor efficient AI in agriculture, holding potential for broader applications.\n  Code: https://github.com/parikshit14/QNN-for-weed\n", "date": "2023-10-29", "categories": []}
{"id": "2310.16991", "title": "An Efficient Deep Learning-based approach for Recognizing Agricultural\n  Pests in the Wild", "abstract": "  One of the biggest challenges that the farmers go through is to fight insect\npests during agricultural product yields. The problem can be solved easily and\navoid economic losses by taking timely preventive measures. This requires\nidentifying insect pests in an easy and effective manner. Most of the insect\nspecies have similarities between them. Without proper help from the\nagriculturist academician it is very challenging for the farmers to identify\nthe crop pests accurately. To address this issue we have done extensive\nexperiments considering different methods to find out the best method among\nall. This paper presents a detailed overview of the experiments done on mainly\na robust dataset named IP102 including transfer learning with finetuning,\nattention mechanism and custom architecture. Some example from another dataset\nD0 is also shown to show robustness of our experimented techniques.\n", "date": "2023-10-25", "categories": []}
{"id": "2308.05407", "title": "A Comparative Assessment of Multi-view fusion learning for Crop\n  Classification", "abstract": "  With a rapidly increasing amount and diversity of remote sensing (RS) data\nsources, there is a strong need for multi-view learning modeling. This is a\ncomplex task when considering the differences in resolution, magnitude, and\nnoise of RS data. The typical approach for merging multiple RS sources has been\ninput-level fusion, but other - more advanced - fusion strategies may\noutperform this traditional approach. This work assesses different fusion\nstrategies for crop classification in the CropHarvest dataset. The fusion\nmethods proposed in this work outperform models based on individual views and\nprevious fusion methods. We do not find one single fusion method that\nconsistently outperforms all other approaches. Instead, we present a comparison\nof multi-view fusion methods for three different datasets and show that,\ndepending on the test region, different methods obtain the best performance.\nDespite this, we suggest a preliminary criterion for the selection of fusion\nmethods.\n", "date": "2023-08-10", "categories": []}
{"id": "2310.12787", "title": "DT/MARS-CycleGAN: Improved Object Detection for MARS Phenotyping Robot", "abstract": "  Robotic crop phenotyping has emerged as a key technology to assess crops'\nmorphological and physiological traits at scale. These phenotypical\nmeasurements are essential for developing new crop varieties with the aim of\nincreasing productivity and dealing with environmental challenges such as\nclimate change. However, developing and deploying crop phenotyping robots face\nmany challenges such as complex and variable crop shapes that complicate\nrobotic object detection, dynamic and unstructured environments that baffle\nrobotic control, and real-time computing and managing big data that challenge\nrobotic hardware/software. This work specifically tackles the first challenge\nby proposing a novel Digital-Twin(DT)MARS-CycleGAN model for image augmentation\nto improve our Modular Agricultural Robotic System (MARS)'s crop object\ndetection from complex and variable backgrounds. Our core idea is that in\naddition to the cycle consistency losses in the CycleGAN model, we designed and\nenforced a new DT-MARS loss in the deep learning model to penalize the\ninconsistency between real crop images captured by MARS and synthesized images\nsensed by DT MARS. Therefore, the generated synthesized crop images closely\nmimic real images in terms of realism, and they are employed to fine-tune\nobject detectors such as YOLOv8. Extensive experiments demonstrated that our\nnew DT/MARS-CycleGAN framework significantly boosts our MARS' crop object/row\ndetector's performance, contributing to the field of robotic crop phenotyping.\n", "date": "2023-10-19", "categories": []}
{"id": "2310.13201", "title": "Identification of Abnormality in Maize Plants From UAV Images Using Deep\n  Learning Approaches", "abstract": "  Early identification of abnormalities in plants is an important task for\nensuring proper growth and achieving high yields from crops. Precision\nagriculture can significantly benefit from modern computer vision tools to make\nfarming strategies addressing these issues efficient and effective. As farming\nlands are typically quite large, farmers have to manually check vast areas to\ndetermine the status of the plants and apply proper treatments. In this work,\nwe consider the problem of automatically identifying abnormal regions in maize\nplants from images captured by a UAV. Using deep learning techniques, we have\ndeveloped a methodology which can detect different levels of abnormality (i.e.,\nlow, medium, high or no abnormality) in maize plants independently of their\ngrowth stage. The primary goal is to identify anomalies at the earliest\npossible stage in order to maximize the effectiveness of potential treatments.\nAt the same time, the proposed system can provide valuable information to human\nannotators for ground truth data collection by helping them to focus their\nattention on a much smaller set of images only. We have experimented with two\ndifferent but complimentary approaches, the first considering abnormality\ndetection as a classification problem and the second considering it as a\nregression problem. Both approaches can be generalized to different types of\nabnormalities and do not make any assumption about the abnormality occurring at\nan early plant growth stage which might be easier to detect due to the plants\nbeing smaller and easier to separate. As a case study, we have considered a\npublicly available data set which exhibits mostly Nitrogen deficiency in maize\nplants of various growth stages. We are reporting promising preliminary results\nwith an 88.89\\% detection accuracy of low abnormality and 100\\% detection\naccuracy of no abnormality.\n", "date": "2023-10-20", "categories": []}
{"id": "2310.12465", "title": "WeedCLR: Weed Contrastive Learning through Visual Representations with\n  Class-Optimized Loss in Long-Tailed Datasets", "abstract": "  Image classification is a crucial task in modern weed management and crop\nintervention technologies. However, the limited size, diversity, and balance of\nexisting weed datasets hinder the development of deep learning models for\ngeneralizable weed identification. In addition, the expensive labelling\nrequirements of mainstream fully-supervised weed classifiers make them cost-\nand time-prohibitive to deploy widely, for new weed species, and in\nsite-specific weed management. This paper proposes a novel method for Weed\nContrastive Learning through visual Representations (WeedCLR), that uses\nclass-optimized loss with Von Neumann Entropy of deep representation for weed\nclassification in long-tailed datasets. WeedCLR leverages self-supervised\nlearning to learn rich and robust visual features without any labels and\napplies a class-optimized loss function to address the class imbalance problem\nin long-tailed datasets. WeedCLR is evaluated on two public weed datasets:\nCottonWeedID15, containing 15 weed species, and DeepWeeds, containing 8 weed\nspecies. WeedCLR achieves an average accuracy improvement of 4.3\\% on\nCottonWeedID15 and 5.6\\% on DeepWeeds over previous methods. It also\ndemonstrates better generalization ability and robustness to different\nenvironmental conditions than existing methods without the need for expensive\nand time-consuming human annotations. These significant improvements make\nWeedCLR an effective tool for weed classification in long-tailed datasets and\nallows for more rapid and widespread deployment of site-specific weed\nmanagement and crop intervention technologies.\n", "date": "2023-10-19", "categories": []}
{"id": "2310.09347", "title": "Efficient Apple Maturity and Damage Assessment: A Lightweight Detection\n  Model with GAN and Attention Mechanism", "abstract": "  This study proposes a method based on lightweight convolutional neural\nnetworks (CNN) and generative adversarial networks (GAN) for apple ripeness and\ndamage level detection tasks. Initially, a lightweight CNN model is designed by\noptimizing the model's depth and width, as well as employing advanced model\ncompression techniques, successfully reducing the model's parameter and\ncomputational requirements, thus enhancing real-time performance in practical\napplications. Simultaneously, attention mechanisms are introduced, dynamically\nadjusting the importance of different feature layers to improve the performance\nin object detection tasks. To address the issues of sample imbalance and\ninsufficient sample size, GANs are used to generate realistic apple images,\nexpanding the training dataset and enhancing the model's recognition capability\nwhen faced with apples of varying ripeness and damage levels. Furthermore, by\napplying the object detection network for damage location annotation on damaged\napples, the accuracy of damage level detection is improved, providing a more\nprecise basis for decision-making. Experimental results show that in apple\nripeness grading detection, the proposed model achieves 95.6\\%, 93.8\\%, 95.0\\%,\nand 56.5 in precision, recall, accuracy, and FPS, respectively. In apple damage\nlevel detection, the proposed model reaches 95.3\\%, 93.7\\%, and 94.5\\% in\nprecision, recall, and mAP, respectively. In both tasks, the proposed method\noutperforms other mainstream models, demonstrating the excellent performance\nand high practical value of the proposed method in apple ripeness and damage\nlevel detection tasks.\n", "date": "2023-10-13", "categories": []}
{"id": "2210.11318", "title": "A Survey of Computer Vision Technologies In Urban and\n  Controlled-environment Agriculture", "abstract": "  In the evolution of agriculture to its next stage, Agriculture 5.0,\nartificial intelligence will play a central role. Controlled-environment\nagriculture, or CEA, is a special form of urban and suburban agricultural\npractice that offers numerous economic, environmental, and social benefits,\nincluding shorter transportation routes to population centers, reduced\nenvironmental impact, and increased productivity. Due to its ability to control\nenvironmental factors, CEA couples well with computer vision (CV) in the\nadoption of real-time monitoring of the plant conditions and autonomous\ncultivation and harvesting. The objective of this paper is to familiarize CV\nresearchers with agricultural applications and agricultural practitioners with\nthe solutions offered by CV. We identify five major CV applications in CEA,\nanalyze their requirements and motivation, and survey the state of the art as\nreflected in 68 technical papers using deep learning methods. In addition, we\ndiscuss five key subareas of computer vision and how they related to these CEA\nproblems, as well as eleven vision-based CEA datasets. We hope the survey will\nhelp researchers quickly gain a bird-eye view of the striving research area and\nwill spark inspiration for new research and development.\n", "date": "2022-10-20", "categories": []}
{"id": "2308.03998", "title": "Real-time Strawberry Detection Based on Improved YOLOv5s Architecture\n  for Robotic Harvesting in open-field environment", "abstract": "  This study proposed a YOLOv5-based custom object detection model to detect\nstrawberries in an outdoor environment. The original architecture of the\nYOLOv5s was modified by replacing the C3 module with the C2f module in the\nbackbone network, which provided a better feature gradient flow. Secondly, the\nSpatial Pyramid Pooling Fast in the final layer of the backbone network of\nYOLOv5s was combined with Cross Stage Partial Net to improve the generalization\nability over the strawberry dataset in this study. The proposed architecture\nwas named YOLOv5s-Straw. The RGB images dataset of the strawberry canopy with\nthree maturity classes (immature, nearly mature, and mature) was collected in\nopen-field environment and augmented through a series of operations including\nbrightness reduction, brightness increase, and noise adding. To verify the\nsuperiority of the proposed method for strawberry detection in open-field\nenvironment, four competitive detection models (YOLOv3-tiny, YOLOv5s,\nYOLOv5s-C2f, and YOLOv8s) were trained, and tested under the same computational\nenvironment and compared with YOLOv5s-Straw. The results showed that the\nhighest mean average precision of 80.3% was achieved using the proposed\narchitecture whereas the same was achieved with YOLOv3-tiny, YOLOv5s,\nYOLOv5s-C2f, and YOLOv8s were 73.4%, 77.8%, 79.8%, 79.3%, respectively.\nSpecifically, the average precision of YOLOv5s-Straw was 82.1% in the immature\nclass, 73.5% in the nearly mature class, and 86.6% in the mature class, which\nwere 2.3% and 3.7%, respectively, higher than that of the latest YOLOv8s. The\nmodel included 8.6*10^6 network parameters with an inference speed of 18ms per\nimage while the inference speed of YOLOv8s had a slower inference speed of\n21.0ms and heavy parameters of 11.1*10^6, which indicates that the proposed\nmodel is fast enough for real time strawberry detection and localization for\nthe robotic picking.\n", "date": "2023-08-08", "categories": []}
{"id": "2310.05929", "title": "Deep Learning based Tomato Disease Detection and Remedy Suggestions\n  using Mobile Application", "abstract": "  We have developed a comprehensive computer system to assist farmers who\npractice traditional farming methods and have limited access to agricultural\nexperts for addressing crop diseases. Our system utilizes artificial\nintelligence (AI) to identify and provide remedies for vegetable diseases. To\nensure ease of use, we have created a mobile application that offers a\nuser-friendly interface, allowing farmers to inquire about vegetable diseases\nand receive suitable solutions in their local language. The developed system\ncan be utilized by any farmer with a basic understanding of a smartphone.\nSpecifically, we have designed an AI-enabled mobile application for identifying\nand suggesting remedies for vegetable diseases, focusing on tomato diseases to\nbenefit the local farming community in Nepal. Our system employs\nstate-of-the-art object detection methodology, namely You Only Look Once\n(YOLO), to detect tomato diseases. The detected information is then relayed to\nthe mobile application, which provides remedy suggestions guided by domain\nexperts. In order to train our system effectively, we curated a dataset\nconsisting of ten classes of tomato diseases. We utilized various data\naugmentation methods to address overfitting and trained a YOLOv5 object\ndetector. The proposed method achieved a mean average precision of 0.76 and\noffers an efficient mobile interface for interacting with the AI system. While\nour system is currently in the development phase, we are actively working\ntowards enhancing its robustness and real-time usability by accumulating more\ntraining samples.\n", "date": "2023-08-16", "categories": []}
{"id": "2310.06582", "title": "Hierarchical Mask2Former: Panoptic Segmentation of Crops, Weeds and\n  Leaves", "abstract": "  Advancements in machine vision that enable detailed inferences to be made\nfrom images have the potential to transform many sectors including agriculture.\nPrecision agriculture, where data analysis enables interventions to be\nprecisely targeted, has many possible applications. Precision spraying, for\nexample, can limit the application of herbicide only to weeds, or limit the\napplication of fertiliser only to undernourished crops, instead of spraying the\nentire field. The approach promises to maximise yields, whilst minimising\nresource use and harms to the surrounding environment. To this end, we propose\na hierarchical panoptic segmentation method to simultaneously identify\nindicators of plant growth and locate weeds within an image. We adapt\nMask2Former, a state-of-the-art architecture for panoptic segmentation, to\npredict crop, weed and leaf masks. We achieve a PQ{\\dag} of 75.99.\nAdditionally, we explore approaches to make the architecture more compact and\ntherefore more suitable for time and compute constrained applications. With our\nmore compact architecture, inference is up to 60% faster and the reduction in\nPQ{\\dag} is less than 1%.\n", "date": "2023-10-10", "categories": []}
{"id": "2310.05425", "title": "Divide and Ensemble: Progressively Learning for the Unknown", "abstract": "  In the wheat nutrient deficiencies classification challenge, we present the\nDividE and EnseMble (DEEM) method for progressive test data predictions. We\nfind that (1) test images are provided in the challenge; (2) samples are\nequipped with their collection dates; (3) the samples of different dates show\nnotable discrepancies. Based on the findings, we partition the dataset into\ndiscrete groups by the dates and train models on each divided group. We then\nadopt the pseudo-labeling approach to label the test data and incorporate those\nwith high confidence into the training set. In pseudo-labeling, we leverage\nmodels ensemble with different architectures to enhance the reliability of\npredictions. The pseudo-labeling and ensembled model training are iteratively\nconducted until all test samples are labeled. Finally, the separated models for\neach group are unified to obtain the model for the whole dataset. Our method\nachieves an average of 93.6\\% Top-1 test accuracy~(94.0\\% on WW2020 and 93.2\\%\non WR2021) and wins the 1$st$ place in the Deep Nutrient Deficiency\nChallenge~\\footnote{https://cvppa2023.github.io/challenges/}.\n", "date": "2023-10-09", "categories": []}
{"id": "2208.04611", "title": "Generative models-based data labeling for deep networks regression:\n  application to seed maturity estimation from UAV multispectral images", "abstract": "  Monitoring seed maturity is an increasing challenge in agriculture due to\nclimate change and more restrictive practices. Seeds monitoring in the field is\nessential to optimize the farming process and to guarantee yield quality\nthrough high germination. Traditional methods are based on limited sampling in\nthe field and analysis in laboratory. Moreover, they are time consuming and\nonly allow monitoring sub-sections of the crop field. This leads to a lack of\naccuracy on the condition of the crop as a whole due to intra-field\nheterogeneities. Multispectral imagery by UAV allows uniform scan of fields and\nbetter capture of crop maturity information. On the other hand, deep learning\nmethods have shown tremendous potential in estimating agronomic parameters,\nespecially maturity. However, they require large labeled datasets. Although\nlarge sets of aerial images are available, labeling them with ground truth is a\ntedious, if not impossible task. In this paper, we propose a method for\nestimating parsley seed maturity using multispectral UAV imagery, with a new\napproach for automatic data labeling. This approach is based on parametric and\nnon-parametric models to provide weak labels. We also consider the data\nacquisition protocol and the performance evaluation of the different steps of\nthe method. Results show good performance, and the non-parametric kernel\ndensity estimator model can improve neural network generalization when used as\na labeling method, leading to more robust and better performing deep neural\nmodels.\n", "date": "2022-08-09", "categories": []}
{"id": "2209.02797", "title": "Fusion of Satellite Images and Weather Data with Transformer Networks\n  for Downy Mildew Disease Detection", "abstract": "  Crop diseases significantly affect the quantity and quality of agricultural\nproduction. In a context where the goal of precision agriculture is to minimize\nor even avoid the use of pesticides, weather and remote sensing data with deep\nlearning can play a pivotal role in detecting crop diseases, allowing localized\ntreatment of crops. However, combining heterogeneous data such as weather and\nimages remains a hot topic and challenging task. Recent developments in\ntransformer architectures have shown the possibility of fusion of data from\ndifferent domains, for instance text-image. The current trend is to custom only\none transformer to create a multimodal fusion model. Conversely, we propose a\nnew approach to realize data fusion using three transformers. In this paper, we\nfirst solved the missing satellite images problem, by interpolating them with a\nConvLSTM model. Then, proposed a multimodal fusion architecture that jointly\nlearns to process visual and weather information. The architecture is built\nfrom three main components, a Vision Transformer and two transformer-encoders,\nallowing to fuse both image and weather modalities. The results of the proposed\nmethod are promising achieving 97\\% overall accuracy.\n", "date": "2022-09-06", "categories": []}
{"id": "2309.16257", "title": "Nondestructive chicken egg fertility detection using CNN-transfer\n  learning algorithms", "abstract": "  This study explored the application of CNN-Transfer Learning for\nnondestructive chicken egg fertility detection for precision poultry hatchery\npractices. Four models, VGG16, ResNet50, InceptionNet, and MobileNet, were\ntrained and evaluated on a dataset (200 single egg images) using augmented\nimages (rotation, flip, scale, translation, and reflection). Although the\ntraining results demonstrated that all models achieved high accuracy,\nindicating their ability to accurately learn and classify chicken eggs'\nfertility state, when evaluated on the testing set, variations in accuracy and\nperformance were observed. InceptionNet exhibited the best overall performance,\naccurately classifying fertile and non-fertile eggs. It demonstrated excellent\nperformance in both training and testing sets in all parameters of the\nevaluation metrics. In testing set, it achieved an accuracy of 0.98, a\nsensitivity of 1 for detecting fertile eggs, and a specificity of 0.96 for\nidentifying non-fertile eggs. The higher performance is attributed to its\nunique architecture efficiently capturing features at different scales leading\nto improved accuracy and robustness. Further optimization and fine-tuning of\nthe models might necessary to address the limitations in accurately detecting\nfertile and non-fertile eggs in case of other models. This study highlighted\nthe potential of CNN-Transfer Learning for nondestructive fertility detection\nand emphasizes the need for further research to enhance the models'\ncapabilities and ensure accurate classification.\n", "date": "2023-09-28", "categories": []}
{"id": "2309.13578", "title": "A SAM-based Solution for Hierarchical Panoptic Segmentation of Crops and\n  Weeds Competition", "abstract": "  Panoptic segmentation in agriculture is an advanced computer vision technique\nthat provides a comprehensive understanding of field composition. It\nfacilitates various tasks such as crop and weed segmentation, plant panoptic\nsegmentation, and leaf instance segmentation, all aimed at addressing\nchallenges in agriculture. Exploring the application of panoptic segmentation\nin agriculture, the 8th Workshop on Computer Vision in Plant Phenotyping and\nAgriculture (CVPPA) hosted the challenge of hierarchical panoptic segmentation\nof crops and weeds using the PhenoBench dataset. To tackle the tasks presented\nin this competition, we propose an approach that combines the effectiveness of\nthe Segment AnyThing Model (SAM) for instance segmentation with prompt input\nfrom object detection models. Specifically, we integrated two notable\napproaches in object detection, namely DINO and YOLO-v8. Our best-performing\nmodel achieved a PQ+ score of 81.33 based on the evaluation metrics of the\ncompetition.\n", "date": "2023-09-24", "categories": []}
{"id": "2309.08739", "title": "Concept explainability for plant diseases classification", "abstract": "  Plant diseases remain a considerable threat to food security and agricultural\nsustainability. Rapid and early identification of these diseases has become a\nsignificant concern motivating several studies to rely on the increasing global\ndigitalization and the recent advances in computer vision based on deep\nlearning. In fact, plant disease classification based on deep convolutional\nneural networks has shown impressive performance. However, these methods have\nyet to be adopted globally due to concerns regarding their robustness,\ntransparency, and the lack of explainability compared with their human experts\ncounterparts. Methods such as saliency-based approaches associating the network\noutput to perturbations of the input pixels have been proposed to give insights\ninto these algorithms. Still, they are not easily comprehensible and not\nintuitive for human users and are threatened by bias. In this work, we deploy a\nmethod called Testing with Concept Activation Vectors (TCAV) that shifts the\nfocus from pixels to user-defined concepts. To the best of our knowledge, our\npaper is the first to employ this method in the field of plant disease\nclassification. Important concepts such as color, texture and disease related\nconcepts were analyzed. The results suggest that concept-based explanation\nmethods can significantly benefit automated plant disease identification.\n", "date": "2023-09-15", "categories": []}
{"id": "2309.07515", "title": "Dhan-Shomadhan: A Dataset of Rice Leaf Disease Classification for\n  Bangladeshi Local Rice", "abstract": "  This dataset represents almost all the harmful diseases for rice in\nBangladesh. This dataset consists of 1106 image of five harmful diseases called\nBrown Spot, Leaf Scaled, Rice Blast, Rice Turngo, Steath Blight in two\ndifferent background variation named field background picture and white\nbackground picture. Two different background variation helps the dataset to\nperform more accurately so that the user can use this data for field use as\nwell as white background for decision making. The data is collected from rice\nfield of Dhaka Division. This dataset can use for rice leaf diseases\nclassification, diseases detection using Computer Vision and Pattern\nRecognition for different rice leaf disease.\n", "date": "2023-09-14", "categories": []}
{"id": "2303.09063", "title": "Plant Disease Detection using Region-Based Convolutional Neural Network", "abstract": "  Agriculture plays an important role in the food and economy of Bangladesh.\nThe rapid growth of population over the years also has increased the demand for\nfood production. One of the major reasons behind low crop production is\nnumerous bacteria, virus and fungal plant diseases. Early detection of plant\ndiseases and proper usage of pesticides and fertilizers are vital for\npreventing the diseases and boost the yield. Most of the farmers use\ngeneralized pesticides and fertilizers in the entire fields without\nspecifically knowing the condition of the plants. Thus the production cost\noftentimes increases, and, not only that, sometimes this becomes detrimental to\nthe yield. Deep Learning models are found to be very effective to automatically\ndetect plant diseases from images of plants, thereby reducing the need for\nhuman specialists. This paper aims at building a lightweight deep learning\nmodel for predicting leaf disease in tomato plants. By modifying the\nregion-based convolutional neural network, we design an efficient and effective\nmodel that demonstrates satisfactory empirical performance on a benchmark\ndataset. Our proposed model can easily be deployed in a larger system where\ndrones take images of leaves and these images will be fed into our model to\nknow the health condition.\n", "date": "2023-03-16", "categories": []}
{"id": "2309.00974", "title": "Deep-Learning Framework for Optimal Selection of Soil Sampling Sites", "abstract": "  This work leverages the recent advancements of deep learning in image\nprocessing to find optimal locations that present the important characteristics\nof a field. The data for training are collected at different fields in local\nfarms with five features: aspect, flow accumulation, slope, NDVI (normalized\ndifference vegetation index), and yield. The soil sampling dataset is\nchallenging because the ground truth is highly imbalanced binary images.\nTherefore, we approached the problem with two methods, the first approach\ninvolves utilizing a state-of-the-art model with the convolutional neural\nnetwork (CNN) backbone, while the second is to innovate a deep-learning design\ngrounded in the concepts of transformer and self-attention. Our framework is\nconstructed with an encoder-decoder architecture with the self-attention\nmechanism as the backbone. In the encoder, the self-attention mechanism is the\nkey feature extractor, which produces feature maps. In the decoder, we\nintroduce atrous convolution networks to concatenate, fuse the extracted\nfeatures, and then export the optimal locations for soil sampling. Currently,\nthe model has achieved impressive results on the testing dataset, with a mean\naccuracy of 99.52%, a mean Intersection over Union (IoU) of 57.35%, and a mean\nDice Coefficient of 71.47%, while the performance metrics of the\nstate-of-the-art CNN-based model are 66.08%, 3.85%, and 1.98%, respectively.\nThis indicates that our proposed model outperforms the CNN-based method on the\nsoil-sampling dataset. To the best of our knowledge, our work is the first to\nprovide a soil-sampling dataset with multiple attributes and leverage deep\nlearning techniques to enable the automatic selection of soil-sampling sites.\nThis work lays a foundation for novel applications of data science and\nmachine-learning technologies to solve other emerging agricultural problems.\n", "date": "2023-09-02", "categories": []}
{"id": "2309.00028", "title": "Vision-Based Cranberry Crop Ripening Assessment", "abstract": "  Agricultural domains are being transformed by recent advances in AI and\ncomputer vision that support quantitative visual evaluation. Using drone\nimaging, we develop a framework for characterizing the ripening process of\ncranberry crops. Our method consists of drone-based time-series collection over\na cranberry growing season, photometric calibration for albedo recovery from\npixels, and berry segmentation with semi-supervised deep learning networks\nusing point-click annotations. By extracting time-series berry albedo\nmeasurements, we evaluate four different varieties of cranberries and provide a\nquantification of their ripening rates. Such quantification has practical\nimplications for 1) assessing real-time overheating risks for cranberry bogs;\n2) large scale comparisons of progeny in crop breeding; 3) detecting disease by\nlooking for ripening pattern outliers. This work is the first of its kind in\nquantitative evaluation of ripening using computer vision methods and has\nimpact beyond cranberry crops including wine grapes, olives, blueberries, and\nmaize.\n", "date": "2023-08-31", "categories": []}
{"id": "2308.08271", "title": "Detecting Olives with Synthetic or Real Data? Olive the Above", "abstract": "  Modern robotics has enabled the advancement in yield estimation for precision\nagriculture. However, when applied to the olive industry, the high variation of\nolive colors and their similarity to the background leaf canopy presents a\nchallenge. Labeling several thousands of very dense olive grove images for\nsegmentation is a labor-intensive task. This paper presents a novel approach to\ndetecting olives without the need to manually label data. In this work, we\npresent the world's first olive detection dataset comprised of synthetic and\nreal olive tree images. This is accomplished by generating an auto-labeled\nphotorealistic 3D model of an olive tree. Its geometry is then simplified for\nlightweight rendering purposes. In addition, experiments are conducted with a\nmix of synthetically generated and real images, yielding an improvement of up\nto 66% compared to when only using a small sample of real data. When access to\nreal, human-labeled data is limited, a combination of mostly synthetic data and\na small amount of real data can enhance olive detection.\n", "date": "2023-08-16", "categories": []}
{"id": "2308.05881", "title": "Aphid Cluster Recognition and Detection in the Wild Using Deep Learning\n  Models", "abstract": "  Aphid infestation poses a significant threat to crop production, rural\ncommunities, and global food security. While chemical pest control is crucial\nfor maximizing yields, applying chemicals across entire fields is both\nenvironmentally unsustainable and costly. Hence, precise localization and\nmanagement of aphids are essential for targeted pesticide application. The\npaper primarily focuses on using deep learning models for detecting aphid\nclusters. We propose a novel approach for estimating infection levels by\ndetecting aphid clusters. To facilitate this research, we have captured a\nlarge-scale dataset from sorghum fields, manually selected 5,447 images\ncontaining aphids, and annotated each individual aphid cluster within these\nimages. To facilitate the use of machine learning models, we further process\nthe images by cropping them into patches, resulting in a labeled dataset\ncomprising 151,380 image patches. Then, we implemented and compared the\nperformance of four state-of-the-art object detection models (VFNet, GFLV2,\nPAA, and ATSS) on the aphid dataset. Extensive experimental results show that\nall models yield stable similar performance in terms of average precision and\nrecall. We then propose to merge close neighboring clusters and remove tiny\nclusters caused by cropping, and the performance is further boosted by around\n17%. The study demonstrates the feasibility of automatically detecting and\nmanaging insects using machine learning models. The labeled dataset will be\nmade openly available to the research community.\n", "date": "2023-08-10", "categories": []}
{"id": "2209.04278", "title": "Deep learning-based Crop Row Detection for Infield Navigation of\n  Agri-Robots", "abstract": "  Autonomous navigation in agricultural environments is challenged by varying\nfield conditions that arise in arable fields. State-of-the-art solutions for\nautonomous navigation in such environments require expensive hardware such as\nRTK-GNSS. This paper presents a robust crop row detection algorithm that\nwithstands such field variations using inexpensive cameras. Existing datasets\nfor crop row detection does not represent all the possible field variations. A\ndataset of sugar beet images was created representing 11 field variations\ncomprised of multiple grow stages, light levels, varying weed densities, curved\ncrop rows and discontinuous crop rows. The proposed pipeline segments the crop\nrows using a deep learning-based method and employs the predicted segmentation\nmask for extraction of the central crop using a novel central crop row\nselection algorithm. The novel crop row detection algorithm was tested for crop\nrow detection performance and the capability of visual servoing along a crop\nrow. The visual servoing-based navigation was tested on a realistic simulation\nscenario with the real ground and plant textures. Our algorithm demonstrated\nrobust vision-based crop row detection in challenging field conditions\noutperforming the baseline.\n", "date": "2022-09-09", "categories": []}
{"id": "2308.05179", "title": "JutePestDetect: An Intelligent Approach for Jute Pest Identification\n  Using Fine-Tuned Transfer Learning", "abstract": "  In certain Asian countries, Jute is one of the primary sources of income and\nGross Domestic Product (GDP) for the agricultural sector. Like many other\ncrops, Jute is prone to pest infestations, and its identification is typically\nmade visually in countries like Bangladesh, India, Myanmar, and China. In\naddition, this method is time-consuming, challenging, and somewhat imprecise,\nwhich poses a substantial financial risk. To address this issue, the study\nproposes a high-performing and resilient transfer learning (TL) based\nJutePestDetect model to identify jute pests at the early stage. Firstly, we\nprepared jute pest dataset containing 17 classes and around 380 photos per pest\nclass, which were evaluated after manual and automatic pre-processing and\ncleaning, such as background removal and resizing. Subsequently, five prominent\npre-trained models -DenseNet201, InceptionV3, MobileNetV2, VGG19, and ResNet50\nwere selected from a previous study to design the JutePestDetect model. Each\nmodel was revised by replacing the classification layer with a global average\npooling layer and incorporating a dropout layer for regularization. To evaluate\nthe models performance, various metrics such as precision, recall, F1 score,\nROC curve, and confusion matrix were employed. These analyses provided\nadditional insights for determining the efficacy of the models. Among them, the\ncustomized regularized DenseNet201-based proposed JutePestDetect model\noutperformed the others, achieving an impressive accuracy of 99%. As a result,\nour proposed method and strategy offer an enhanced approach to pest\nidentification in the case of Jute, which can significantly benefit farmers\nworldwide.\n", "date": "2023-05-28", "categories": []}
{"id": "2304.04963", "title": "PlantDet: A benchmark for Plant Detection in the Three-Rivers-Source\n  Region", "abstract": "  The Three-River-Source region is a highly significant natural reserve in\nChina that harbors a plethora of botanical resources. To meet the practical\nrequirements of botanical research and intelligent plant management, we\nconstruct a dataset for Plant detection in the Three-River-Source region\n(PTRS). It comprises 21 types, 6965 high-resolution images of 2160*3840 pixels,\ncaptured by diverse sensors and platforms, and featuring objects of varying\nshapes and sizes. The PTRS presents us with challenges such as dense occlusion,\nvarying leaf resolutions, and high feature similarity among plants, prompting\nus to develop a novel object detection network named PlantDet. This network\nemploys a window-based efficient self-attention module (ST block) to generate\nrobust feature representation at multiple scales, improving the detection\nefficiency for small and densely-occluded objects. Our experimental results\nvalidate the efficacy of our proposed plant detection benchmark, with a\nprecision of 88.1%, a mean average precision (mAP) of 77.6%, and a higher\nrecall compared to the baseline. Additionally, our method effectively overcomes\nthe issue of missing small objects.\n", "date": "2023-04-11", "categories": []}
{"id": "2208.14508", "title": "Swin-transformer-yolov5 For Real-time Wine Grape Bunch Detection", "abstract": "  In this research, an integrated detection model, Swin-transformer-YOLOv5 or\nSwin-T-YOLOv5, was proposed for real-time wine grape bunch detection to inherit\nthe advantages from both YOLOv5 and Swin-transformer. The research was\nconducted on two different grape varieties of Chardonnay (always white berry\nskin) and Merlot (white or white-red mix berry skin when immature; red when\nmatured) from July to September in 2019. To verify the superiority of\nSwin-T-YOLOv5, its performance was compared against several commonly\nused/competitive object detectors, including Faster R-CNN, YOLOv3, YOLOv4, and\nYOLOv5. All models were assessed under different test conditions, including two\ndifferent weather conditions (sunny and cloudy), two different berry maturity\nstages (immature and mature), and three different sunlight\ndirections/intensities (morning, noon, and afternoon) for a comprehensive\ncomparison. Additionally, the predicted number of grape bunches by\nSwin-T-YOLOv5 was further compared with ground truth values, including both\nin-field manual counting and manual labeling during the annotation process.\nResults showed that the proposed Swin-T-YOLOv5 outperformed all other studied\nmodels for grape bunch detection, with up to 97% of mean Average Precision\n(mAP) and 0.89 of F1-score when the weather was cloudy. This mAP was\napproximately 44%, 18%, 14%, and 4% greater than Faster R-CNN, YOLOv3, YOLOv4,\nand YOLOv5, respectively. Swin-T-YOLOv5 achieved its lowest mAP (90%) and\nF1-score (0.82) when detecting immature berries, where the mAP was\napproximately 40%, 5%, 3%, and 1% greater than the same. Furthermore,\nSwin-T-YOLOv5 performed better on Chardonnay variety with achieved up to 0.91\nof R2 and 2.36 root mean square error (RMSE) when comparing the predictions\nwith ground truth. However, it underperformed on Merlot variety with achieved\nonly up to 0.70 of R2 and 3.30 of RMSE.\n", "date": "2022-08-30", "categories": []}
{"id": "2305.15149", "title": "Reliability Scores from Saliency Map Clusters for Improved Image-based\n  Harvest-Readiness Prediction in Cauliflower", "abstract": "  Cauliflower is a hand-harvested crop that must fulfill high-quality standards\nin sales making the timing of harvest important. However, accurately\ndetermining harvest-readiness can be challenging due to the cauliflower head\nbeing covered by its canopy. While deep learning enables automated\nharvest-readiness estimation, errors can occur due to field-variability and\nlimited training data. In this paper, we analyze the reliability of a\nharvest-readiness classifier with interpretable machine learning. By\nidentifying clusters of saliency maps, we derive reliability scores for each\nclassification result using knowledge about the domain and the image\nproperties. For unseen data, the reliability can be used to (i) inform farmers\nto improve their decision-making and (ii) increase the model prediction\naccuracy. Using RGB images of single cauliflower plants at different\ndevelopmental stages from the GrowliFlower dataset, we investigate various\nsaliency mapping approaches and find that they result in different quality of\nreliability scores. With the most suitable interpretation tool, we adjust the\nclassification result and achieve a 15.72% improvement of the overall accuracy\nto 88.14% and a 15.44% improvement of the average class accuracy to 88.52% for\nthe GrowliFlower dataset.\n", "date": "2023-05-24", "categories": []}
{"id": "2308.03766", "title": "AMaizeD: An End to End Pipeline for Automatic Maize Disease Detection", "abstract": "  This research paper presents AMaizeD: An End to End Pipeline for Automatic\nMaize Disease Detection, an automated framework for early detection of diseases\nin maize crops using multispectral imagery obtained from drones. A custom\nhand-collected dataset focusing specifically on maize crops was meticulously\ngathered by expert researchers and agronomists. The dataset encompasses a\ndiverse range of maize varieties, cultivation practices, and environmental\nconditions, capturing various stages of maize growth and disease progression.\nBy leveraging multispectral imagery, the framework benefits from improved\nspectral resolution and increased sensitivity to subtle changes in plant\nhealth. The proposed framework employs a combination of convolutional neural\nnetworks (CNNs) as feature extractors and segmentation techniques to identify\nboth the maize plants and their associated diseases. Experimental results\ndemonstrate the effectiveness of the framework in detecting a range of maize\ndiseases, including powdery mildew, anthracnose, and leaf blight. The framework\nachieves state-of-the-art performance on the custom hand-collected dataset and\ncontributes to the field of automated disease detection in agriculture,\noffering a practical solution for early identification of diseases in maize\ncrops advanced machine learning techniques and deep learning architectures.\n", "date": "2023-07-23", "categories": []}
{"id": "2308.03670", "title": "Improving FHB Screening in Wheat Breeding Using an Efficient Transformer\n  Model", "abstract": "  Fusarium head blight is a devastating disease that causes significant\neconomic losses annually on small grains. Efficiency, accuracy, and timely\ndetection of FHB in the resistance screening are critical for wheat and barley\nbreeding programs. In recent years, various image processing techniques have\nbeen developed using supervised machine learning algorithms for the early\ndetection of FHB. The state-of-the-art convolutional neural network-based\nmethods, such as U-Net, employ a series of encoding blocks to create a local\nrepresentation and a series of decoding blocks to capture the semantic\nrelations. However, these methods are not often capable of long-range modeling\ndependencies inside the input data, and their ability to model multi-scale\nobjects with significant variations in texture and shape is limited. Vision\ntransformers as alternative architectures with innate global self-attention\nmechanisms for sequence-to-sequence prediction, due to insufficient low-level\ndetails, may also limit localization capabilities. To overcome these\nlimitations, a new Context Bridge is proposed to integrate the local\nrepresentation capability of the U-Net network in the transformer model. In\naddition, the standard attention mechanism of the original transformer is\nreplaced with Efficient Self-attention, which is less complicated than other\nstate-of-the-art methods. To train the proposed network, 12,000 wheat images\nfrom an FHB-inoculated wheat field at the SDSU research farm in Volga, SD, were\ncaptured. In addition to healthy and unhealthy plants, these images encompass\nvarious stages of the disease. A team of expert pathologists annotated the\nimages for training and evaluating the developed model. As a result, the\neffectiveness of the transformer-based method for FHB-disease detection,\nthrough extensive experiments across typical tasks for plant image\nsegmentation, is demonstrated.\n", "date": "2023-08-07", "categories": []}
{"id": "2211.02972", "title": "Inside Out: Transforming Images of Lab-Grown Plants for Machine Learning\n  Applications in Agriculture", "abstract": "  Machine learning tasks often require a significant amount of training data\nfor the resultant network to perform suitably for a given problem in any\ndomain. In agriculture, dataset sizes are further limited by phenotypical\ndifferences between two plants of the same genotype, often as a result of\ndiffering growing conditions. Synthetically-augmented datasets have shown\npromise in improving existing models when real data is not available. In this\npaper, we employ a contrastive unpaired translation (CUT) generative\nadversarial network (GAN) and simple image processing techniques to translate\nindoor plant images to appear as field images. While we train our network to\ntranslate an image containing only a single plant, we show that our method is\neasily extendable to produce multiple-plant field images. Furthermore, we use\nour synthetic multi-plant images to train several YoloV5 nano object detection\nmodels to perform the task of plant detection and measure the accuracy of the\nmodel on real field data images. Including training data generated by the\nCUT-GAN leads to better plant detection performance compared to a network\ntrained solely on real data.\n", "date": "2022-11-05", "categories": []}
{"id": "2307.11706", "title": "3D Skeletonization of Complex Grapevines for Robotic Pruning", "abstract": "  Robotic pruning of dormant grapevines is an area of active research in order\nto promote vine balance and grape quality, but so far robotic efforts have\nlargely focused on planar, simplified vines not representative of commercial\nvineyards. This paper aims to advance the robotic perception capabilities\nnecessary for pruning in denser and more complex vine structures by extending\nplant skeletonization techniques. The proposed pipeline generates skeletal\ngrapevine models that have lower reprojection error and higher connectivity\nthan baseline algorithms. We also show how 3D and skeletal information enables\nprediction accuracy of pruning weight for dense vines surpassing prior work,\nwhere pruning weight is an important vine metric influencing pruning site\nselection.\n", "date": "2023-07-21", "categories": []}
{"id": "2307.10267", "title": "On the Real-Time Semantic Segmentation of Aphid Clusters in the Wild", "abstract": "  Aphid infestations can cause extensive damage to wheat and sorghum fields and\nspread plant viruses, resulting in significant yield losses in agriculture. To\naddress this issue, farmers often rely on chemical pesticides, which are\ninefficiently applied over large areas of fields. As a result, a considerable\namount of pesticide is wasted on areas without pests, while inadequate amounts\nare applied to areas with severe infestations. The paper focuses on the urgent\nneed for an intelligent autonomous system that can locate and spray\ninfestations within complex crop canopies, reducing pesticide use and\nenvironmental impact. We have collected and labeled a large aphid image dataset\nin the field, and propose the use of real-time semantic segmentation models to\nsegment clusters of aphids. A multiscale dataset is generated to allow for\nlearning the clusters at different scales. We compare the segmentation speeds\nand accuracy of four state-of-the-art real-time semantic segmentation models on\nthe aphid cluster dataset, benchmarking them against nonreal-time models. The\nstudy results show the effectiveness of a real-time solution, which can reduce\ninefficient pesticide use and increase crop yields, paving the way towards an\nautonomous pest detection system.\n", "date": "2023-07-17", "categories": []}
{"id": "2307.07863", "title": "Benchmarking the Effectiveness of Classification Algorithms and SVM\n  Kernels for Dry Beans", "abstract": "  Plant breeders and agricultural researchers can increase crop productivity by\nidentifying desirable features, disease resistance, and nutritional content by\nanalysing the Dry Bean dataset. This study analyses and compares different\nSupport Vector Machine (SVM) classification algorithms, namely linear,\npolynomial, and radial basis function (RBF), along with other popular\nclassification algorithms. The analysis is performed on the Dry Bean Dataset,\nwith PCA (Principal Component Analysis) conducted as a preprocessing step for\ndimensionality reduction. The primary evaluation metric used is accuracy, and\nthe RBF SVM kernel algorithm achieves the highest Accuracy of 93.34%, Precision\nof 92.61%, Recall of 92.35% and F1 Score as 91.40%. Along with adept\nvisualization and empirical analysis, this study offers valuable guidance by\nemphasizing the importance of considering different SVM algorithms for complex\nand non-linear structured datasets.\n", "date": "2023-07-15", "categories": []}
{"id": "2307.05929", "title": "A New Dataset and Comparative Study for Aphid Cluster Detection", "abstract": "  Aphids are one of the main threats to crops, rural families, and global food\nsecurity. Chemical pest control is a necessary component of crop production for\nmaximizing yields, however, it is unnecessary to apply the chemical approaches\nto the entire fields in consideration of the environmental pollution and the\ncost. Thus, accurately localizing the aphid and estimating the infestation\nlevel is crucial to the precise local application of pesticides. Aphid\ndetection is very challenging as each individual aphid is really small and all\naphids are crowded together as clusters. In this paper, we propose to estimate\nthe infection level by detecting aphid clusters. We have taken millions of\nimages in the sorghum fields, manually selected 5,447 images that contain\naphids, and annotated each aphid cluster in the image. To use these images for\nmachine learning models, we crop the images into patches and created a labeled\ndataset with over 151,000 image patches. Then, we implement and compare the\nperformance of four state-of-the-art object detection models.\n", "date": "2023-07-12", "categories": []}
{"id": "2301.00363", "title": "Mapping smallholder cashew plantations to inform sustainable tree crop\n  expansion in Benin", "abstract": "  Cashews are grown by over 3 million smallholders in more than 40 countries\nworldwide as a principal source of income. As the third largest cashew producer\nin Africa, Benin has nearly 200,000 smallholder cashew growers contributing 15%\nof the country's national export earnings. However, a lack of information on\nwhere and how cashew trees grow across the country hinders decision-making that\ncould support increased cashew production and poverty alleviation. By\nleveraging 2.4-m Planet Basemaps and 0.5-m aerial imagery, newly developed deep\nlearning algorithms, and large-scale ground truth datasets, we successfully\nproduced the first national map of cashew in Benin and characterized the\nexpansion of cashew plantations between 2015 and 2021. In particular, we\ndeveloped a SpatioTemporal Classification with Attention (STCA) model to map\nthe distribution of cashew plantations, which can fully capture texture\ninformation from discriminative time steps during a growing season. We further\ndeveloped a Clustering Augmented Self-supervised Temporal Classification\n(CASTC) model to distinguish high-density versus low-density cashew plantations\nby automatic feature extraction and optimized clustering. Results show that the\nSTCA model has an overall accuracy over 85% and the CASTC model achieved an\noverall accuracy of 76%. We found that the cashew area in Benin almost doubled\nfrom 2015 to 2021 with 60% of new plantation development coming from cropland\nor fallow land, while encroachment of cashew plantations into protected areas\nhas increased by 55%. Only half of cashew plantations were high-density in\n2021, suggesting high potential for intensification. Our study illustrates the\npower of combining high-resolution remote sensing imagery and state-of-the-art\ndeep learning algorithms to better understand tree crops in the heterogeneous\nsmallholder landscape.\n", "date": "2023-01-01", "categories": []}
{"id": "2307.01767", "title": "Localized Data Work as a Precondition for Data-Centric ML: A Case Study\n  of Full Lifecycle Crop Disease Identification in Ghana", "abstract": "  The Ghana Cashew Disease Identification with Artificial Intelligence (CADI\nAI) project demonstrates the importance of sound data work as a precondition\nfor the delivery of useful, localized datacentric solutions for public good\ntasks such as agricultural productivity and food security. Drone collected data\nand machine learning are utilized to determine crop stressors. Data, model and\nthe final app are developed jointly and made available to local farmers via a\ndesktop application.\n", "date": "2023-07-04", "categories": []}
{"id": "2307.01064", "title": "TomatoDIFF: On-plant Tomato Segmentation with Denoising Diffusion Models", "abstract": "  Artificial intelligence applications enable farmers to optimize crop growth\nand production while reducing costs and environmental impact. Computer\nvision-based algorithms in particular, are commonly used for fruit\nsegmentation, enabling in-depth analysis of the harvest quality and accurate\nyield estimation. In this paper, we propose TomatoDIFF, a novel diffusion-based\nmodel for semantic segmentation of on-plant tomatoes. When evaluated against\nother competitive methods, our model demonstrates state-of-the-art (SOTA)\nperformance, even in challenging environments with highly occluded fruits.\nAdditionally, we introduce Tomatopia, a new, large and challenging dataset of\ngreenhouse tomatoes. The dataset comprises high-resolution RGB-D images and\npixel-level annotations of the fruits.\n", "date": "2023-07-03", "categories": []}
{"id": "2306.16734", "title": "Unified View of Damage leaves Planimetry & Analysis Using Digital Images\n  Processing Techniques", "abstract": "  The detection of leaf diseases in plants generally involves visual\nobservation of patterns appearing on the leaf surface. However, there are many\ndiseases that are distinguished based on very subtle changes in these visually\nobservable patterns. This paper attempts to identify plant leaf diseases using\nimage processing techniques. The focus of this study is on the detection of\ncitrus leaf canker disease. Canker is a bacterial infection of leaves. Symptoms\nof citrus cankers include brown spots on the leaves, often with a watery or\noily appearance. The spots (called lesions in botany) are usually yellow. It is\nsurrounded by a halo of the leaves and is found on both the top and bottom of\nthe leaf. This paper describes various methods that have been used to detect\ncitrus leaf canker disease. The methods used are histogram comparison and\nk-means clustering. Using these methods, citrus canker development was detected\nbased on histograms generated based on leaf patterns. The results thus obtained\ncan be used, after consultation with experts in the field of agriculture, to\nidentify suitable treatments for the processes used.\n", "date": "2023-06-29", "categories": []}
{"id": "2306.16862", "title": "Sustainable Palm Tree Farming: Leveraging IoT and Multi-Modal Data for\n  Early Detection and Mapping of Red Palm Weevil", "abstract": "  The Red Palm Weevil (RPW) is a highly destructive insect causing economic\nlosses and impacting palm tree farming worldwide. This paper proposes an\ninnovative approach for sustainable palm tree farming by utilizing advanced\ntechnologies for the early detection and management of RPW. Our approach\ncombines computer vision, deep learning (DL), the Internet of Things (IoT), and\ngeospatial data to detect and classify RPW-infested palm trees effectively. The\nmain phases include; (1) DL classification using sound data from IoT devices,\n(2) palm tree detection using YOLOv8 on UAV images, and (3) RPW mapping using\ngeospatial data. Our custom DL model achieves 100% precision and recall in\ndetecting and localizing infested palm trees. Integrating geospatial data\nenables the creation of a comprehensive RPW distribution map for efficient\nmonitoring and targeted management strategies. This technology-driven approach\nbenefits agricultural authorities, farmers, and researchers in managing RPW\ninfestations and safeguarding palm tree plantations' productivity.\n", "date": "2023-06-29", "categories": []}
{"id": "2306.13518", "title": "Segmentation and Tracking of Vegetable Plants by Exploiting Vegetable\n  Shape Feature for Precision Spray of Agricultural Robots", "abstract": "  With the increasing deployment of agricultural robots, the traditional manual\nspray of liquid fertilizer and pesticide is gradually being replaced by\nagricultural robots. For robotic precision spray application in vegetable\nfarms, accurate plant phenotyping through instance segmentation and robust\nplant tracking are of great importance and a prerequisite for the following\nspray action. Regarding the robust tracking of vegetable plants, to solve the\nchallenging problem of associating vegetables with similar color and texture in\nconsecutive images, in this paper, a novel method of Multiple Object Tracking\nand Segmentation (MOTS) is proposed for instance segmentation and tracking of\nmultiple vegetable plants. In our approach, contour and blob features are\nextracted to describe unique feature of each individual vegetable, and\nassociate the same vegetables in different images. By assigning a unique ID for\neach vegetable, it ensures the robot to spray each vegetable exactly once,\nwhile traversing along the farm rows. Comprehensive experiments including\nablation studies are conducted, which prove its superior performance over two\nState-Of-The-Art (SOTA) MOTS methods. Compared to the conventional MOTS\nmethods, the proposed method is able to re-identify objects which have gone out\nof the camera field of view and re-appear again using the proposed data\nassociation strategy, which is important to ensure each vegetable be sprayed\nonly once when the robot travels back and forth. Although the method is tested\non lettuce farm, it can be applied to other similar vegetables such as broccoli\nand canola. Both code and the dataset of this paper is publicly released for\nthe benefit of the community: https://github.com/NanH5837/LettuceMOTS.\n", "date": "2023-06-23", "categories": []}
{"id": "2306.13366", "title": "Lesion Detection on Leaves using Class Activation Maps", "abstract": "  Lesion detection on plant leaves is a critical task in plant pathology and\nagricultural research. Identifying lesions enables assessing the severity of\nplant diseases and making informed decisions regarding disease control measures\nand treatment strategies. To detect lesions, there are studies that propose\nwell-known object detectors. However, training object detectors to detect small\nobjects such as lesions can be problematic. In this study, we propose a method\nfor lesion detection on plant leaves utilizing class activation maps generated\nby a ResNet-18 classifier. In the test set, we achieved a 0.45 success rate in\npredicting the locations of lesions in leaves. Our study presents a novel\napproach for lesion detection on plant leaves by utilizing CAMs generated by a\nResNet classifier while eliminating the need for a lesion annotation process.\n", "date": "2023-06-23", "categories": []}
{"id": "2306.12057", "title": "Chili Pepper Disease Diagnosis via Image Reconstruction Using GrabCut\n  and Generative Adversarial Serial Autoencoder", "abstract": "  With the recent development of smart farms, researchers are very interested\nin such fields. In particular, the field of disease diagnosis is the most\nimportant factor. Disease diagnosis belongs to the field of anomaly detection\nand aims to distinguish whether plants or fruits are normal or abnormal. The\nproblem can be solved by binary or multi-classification based on CNN, but it\ncan also be solved by image reconstruction. However, due to the limitation of\nthe performance of image generation, SOTA's methods propose a score calculation\nmethod using a latent vector error. In this paper, we propose a network that\nfocuses on chili peppers and proceeds with background removal through Grabcut.\nIt shows high performance through image-based score calculation method. Due to\nthe difficulty of reconstructing the input image, the difference between the\ninput and output images is large. However, the serial autoencoder proposed in\nthis paper uses the difference between the two fake images except for the\nactual input as a score. We propose a method of generating meaningful images\nusing the GAN structure and classifying three results simultaneously by one\ndiscriminator. The proposed method showed higher performance than previous\nresearches, and image-based scores showed the best performanc\n", "date": "2023-06-21", "categories": []}
{"id": "2210.07879", "title": "Hierarchical Approach for Joint Semantic, Plant Instance, and Leaf\n  Instance Segmentation in the Agricultural Domain", "abstract": "  Plant phenotyping is a central task in agriculture, as it describes plants'\ngrowth stage, development, and other relevant quantities. Robots can help\nautomate this process by accurately estimating plant traits such as the number\nof leaves, leaf area, and the plant size. In this paper, we address the problem\nof joint semantic, plant instance, and leaf instance segmentation of crop\nfields from RGB data. We propose a single convolutional neural network that\naddresses the three tasks simultaneously, exploiting their underlying\nhierarchical structure. We introduce task-specific skip connections, which our\nexperimental evaluation proves to be more beneficial than the usual schemes. We\nalso propose a novel automatic post-processing, which explicitly addresses the\nproblem of spatially close instances, common in the agricultural domain because\nof overlapping leaves. Our architecture simultaneously tackles these problems\njointly in the agricultural context. Previous works either focus on plant or\nleaf segmentation, or do not optimise for semantic segmentation. Results show\nthat our system has superior performance compared to state-of-the-art\napproaches, while having a reduced number of parameters and is operating at\ncamera frame rate.\n", "date": "2022-10-14", "categories": []}
{"id": "2306.06553", "title": "Hinting Pipeline and Multivariate Regression CNN for Maize Kernel\n  Counting on the Ear", "abstract": "  Maize is a highly nutritional cereal widely used for human and animal\nconsumption and also as raw material by the biofuels industries. This\nhighlights the importance of precisely quantifying the corn grain productivity\nin season, helping the commercialization process, operationalization, and\ncritical decision-making. Considering the manual labor cost of counting maize\nkernels, we propose in this work a novel preprocessing pipeline named hinting\nthat guides the attention of the model to the center of the corn kernels and\nenables a deep learning model to deliver better performance, given a picture of\none side of the corn ear. Also, we propose a multivariate CNN regressor that\noutperforms single regression results. Experiments indicated that the proposed\napproach excels the current manual estimates, obtaining MAE of 34.4 and R2 of\n0.74 against 35.38 and 0.72 for the manual estimate, respectively.\n", "date": "2023-06-11", "categories": []}
{"id": "2306.07244", "title": "RB-Dust -- A Reference-based Dataset for Vision-based Dust Removal", "abstract": "  Dust in the agricultural landscape is a significant challenge and influences,\nfor example, the environmental perception of autonomous agricultural machines.\nImage enhancement algorithms can be used to reduce dust. However, these require\ndusty and dust-free images of the same environment for validation. In fact, to\ndate, there is no dataset that we are aware of that addresses this issue.\nTherefore, we present the agriscapes RB-Dust dataset, which is named after its\npurpose of reference-based dust removal. It is not possible to take pictures\nfrom the cabin during tillage, as this would cause shifts in the images.\nBecause of this, we built a setup from which it is possible to take images from\na stationary position close to the passing tractor. The test setup was based on\na half-sided gate through which the tractor could drive. The field tests were\ncarried out on a farm in Bavaria, Germany, during tillage. During the field\ntests, other parameters such as soil moisture and wind speed were controlled,\nas these significantly affect dust development. We validated our dataset with\ncontrast enhancement and image dehazing algorithms and analyzed the\ngeneralizability from recordings from the moving tractor. Finally, we\ndemonstrate the application of dust removal based on a high-level vision task,\nsuch as person classification. Our empirical study confirms the validity of\nRB-Dust for vision-based dust removal in agriculture.\n", "date": "2023-06-12", "categories": []}
{"id": "2306.06080", "title": "Detection of Late Blight Disease in Tomato Leaf Using Image Processing\n  Techniques", "abstract": "  =One of the most frequently farmed crops is the tomato crop. Late blight is\nthe most prevalent tomato disease in the world, and often causes a significant\nreduction in the production of tomato crops. The importance of tomatoes as an\nagricultural product necessitates early detection of late blight. It is\nproduced by the fungus Phytophthora. The earliest signs of late blight on\ntomatoes are unevenly formed, water-soaked lesions on the leaves located on the\nplant canopy's younger leave White cottony growth may appear in humid\nenvironments evident on the undersides of the leaves that have been impacted.\nLesions increase as the disease proceeds, turning the leaves brown to shrivel\nup and die. Using picture segmentation and the Multi-class SVM technique, late\nblight disorder is discovered in this work. Image segmentation is employed for\nseparating damaged areas on leaves, and the Multi-class SVM method is used for\nreliable disease categorization. 30 reputable studies were chosen from a total\nof 2770 recognized papers. The primary goal of this study is to compile\ncutting-edge research that identifies current research trends, problems, and\nprospects for late blight detection. It also looks at current approaches for\napplying image processing to diagnose and detect late blight. A suggested\ntaxonomy for late blight detection has also been provided. In the same way, a\nmodel for the development of the solutions to problems is also presented.\nFinally, the research gaps have been presented in terms of open issues for the\nprovision of future directions in image processing for the researchers.\n", "date": "2023-05-31", "categories": []}
{"id": "2305.14691", "title": "Label-Efficient Learning in Agriculture: A Comprehensive Review", "abstract": "  The past decade has witnessed many great successes of machine learning (ML)\nand deep learning (DL) applications in agricultural systems, including weed\ncontrol, plant disease diagnosis, agricultural robotics, and precision\nlivestock management. Despite tremendous progresses, one downside of such ML/DL\nmodels is that they generally rely on large-scale labeled datasets for\ntraining, and the performance of such models is strongly influenced by the size\nand quality of available labeled data samples. In addition, collecting,\nprocessing, and labeling such large-scale datasets is extremely costly and\ntime-consuming, partially due to the rising cost in human labor. Therefore,\ndeveloping label-efficient ML/DL methods for agricultural applications has\nreceived significant interests among researchers and practitioners. In fact,\nthere are more than 50 papers on developing and applying deep-learning-based\nlabel-efficient techniques to address various agricultural problems since 2016,\nwhich motivates the authors to provide a timely and comprehensive review of\nrecent label-efficient ML/DL methods in agricultural applications. To this end,\nwe first develop a principled taxonomy to organize these methods according to\nthe degree of supervision, including weak supervision (i.e., active learning\nand semi-/weakly- supervised learning), and no supervision (i.e., un-/self-\nsupervised learning), supplemented by representative state-of-the-art\nlabel-efficient ML/DL methods. In addition, a systematic review of various\nagricultural applications exploiting these label-efficient algorithms, such as\nprecision agriculture, plant phenotyping, and postharvest quality assessment,\nis presented. Finally, we discuss the current problems and challenges, as well\nas future research directions. A well-classified paper list can be accessed at\nhttps://github.com/DongChen06/Label-efficient-in-Agriculture.\n", "date": "2023-05-24", "categories": []}
{"id": "2305.13490", "title": "Detection of healthy and diseased crops in drone captured images using\n  Deep Learning", "abstract": "  Monitoring plant health is crucial for maintaining agricultural productivity\nand food safety. Disruptions in the plant's normal state, caused by diseases,\noften interfere with essential plant activities, and timely detection of these\ndiseases can significantly mitigate crop loss. In this study, we propose a deep\nlearning-based approach for efficient detection of plant diseases using\ndrone-captured imagery. A comprehensive database of various plant species,\nexhibiting numerous diseases, was compiled from the Internet and utilized as\nthe training and test dataset. A Convolutional Neural Network (CNN), renowned\nfor its performance in image classification tasks, was employed as our primary\npredictive model. The CNN model, trained on this rich dataset, demonstrated\nsuperior proficiency in crop disease categorization and detection, even under\nchallenging imaging conditions. For field implementation, we deployed a\nprototype drone model equipped with a high-resolution camera for live\nmonitoring of extensive agricultural fields. The captured images served as the\ninput for our trained model, enabling real-time identification of healthy and\ndiseased plants. Our approach promises an efficient and scalable solution for\nimproving crop health monitoring systems.\n", "date": "2023-05-22", "categories": []}
{"id": "2305.09418", "title": "Leaf Only SAM: A Segment Anything Pipeline for Zero-Shot Automated Leaf\n  Segmentation", "abstract": "  Segment Anything Model (SAM) is a new foundation model that can be used as a\nzero-shot object segmentation method with the use of either guide prompts such\nas bounding boxes, polygons, or points. Alternatively, additional post\nprocessing steps can be used to identify objects of interest after segmenting\neverything in an image. Here we present a method using segment anything\ntogether with a series of post processing steps to segment potato leaves,\ncalled Leaf Only SAM. The advantage of this proposed method is that it does not\nrequire any training data to produce its results so has many applications\nacross the field of plant phenotyping where there is limited high quality\nannotated data available. We compare the performance of Leaf Only SAM to a Mask\nR-CNN model which has been fine-tuned on our small novel potato leaf dataset.\nOn the evaluation dataset, Leaf Only SAM finds an average recall of 63.2 and an\naverage precision of 60.3, compared to recall of 78.7 and precision of 74.7 for\nMask R-CNN. Leaf Only SAM does not perform better than the fine-tuned Mask\nR-CNN model on our data, but the SAM based model does not require any extra\ntraining or annotation of our new dataset. This shows there is potential to use\nSAM as a zero-shot classifier with the addition of post processing steps.\n", "date": "2023-05-16", "categories": []}
{"id": "2301.10351", "title": "Few-Shot Learning Enables Population-Scale Analysis of Leaf Traits in\n  Populus trichocarpa", "abstract": "  Plant phenotyping is typically a time-consuming and expensive endeavor,\nrequiring large groups of researchers to meticulously measure biologically\nrelevant plant traits, and is the main bottleneck in understanding plant\nadaptation and the genetic architecture underlying complex traits at population\nscale. In this work, we address these challenges by leveraging few-shot\nlearning with convolutional neural networks (CNNs) to segment the leaf body and\nvisible venation of 2,906 P. trichocarpa leaf images obtained in the field. In\ncontrast to previous methods, our approach (i) does not require experimental or\nimage pre-processing, (ii) uses the raw RGB images at full resolution, and\n(iii) requires very few samples for training (e.g., just eight images for vein\nsegmentation). Traits relating to leaf morphology and vein topology are\nextracted from the resulting segmentations using traditional open-source\nimage-processing tools, validated using real-world physical measurements, and\nused to conduct a genome-wide association study to identify genes controlling\nthe traits. In this way, the current work is designed to provide the plant\nphenotyping community with (i) methods for fast and accurate image-based\nfeature extraction that require minimal training data, and (ii) a new\npopulation-scale data set, including 68 different leaf phenotypes, for domain\nscientists and machine learning researchers. All of the few-shot learning code,\ndata, and results are made publicly available.\n", "date": "2023-01-24", "categories": []}
{"id": "2305.09810", "title": "Semi-Supervised Object Detection for Sorghum Panicles in UAV Imagery", "abstract": "  The sorghum panicle is an important trait related to grain yield and plant\ndevelopment. Detecting and counting sorghum panicles can provide significant\ninformation for plant phenotyping. Current deep-learning-based object detection\nmethods for panicles require a large amount of training data. The data labeling\nis time-consuming and not feasible for real application. In this paper, we\npresent an approach to reduce the amount of training data for sorghum panicle\ndetection via semi-supervised learning. Results show we can achieve similar\nperformance as supervised methods for sorghum panicle detection by only using\n10\\% of original training data.\n", "date": "2023-05-16", "categories": []}
{"id": "2305.10254", "title": "SAM for Poultry Science", "abstract": "  In recent years, the agricultural industry has witnessed significant\nadvancements in artificial intelligence (AI), particularly with the development\nof large-scale foundational models. Among these foundation models, the Segment\nAnything Model (SAM), introduced by Meta AI Research, stands out as a\ngroundbreaking solution for object segmentation tasks. While SAM has shown\nsuccess in various agricultural applications, its potential in the poultry\nindustry, specifically in the context of cage-free hens, remains relatively\nunexplored. This study aims to assess the zero-shot segmentation performance of\nSAM on representative chicken segmentation tasks, including part-based\nsegmentation and the use of infrared thermal images, and to explore\nchicken-tracking tasks by using SAM as a segmentation tool. The results\ndemonstrate SAM's superior performance compared to SegFormer and SETR in both\nwhole and part-based chicken segmentation. SAM-based object tracking also\nprovides valuable data on the behavior and movement patterns of broiler birds.\nThe findings of this study contribute to a better understanding of SAM's\npotential in poultry science and lay the foundation for future advancements in\nchicken segmentation and tracking.\n", "date": "2023-05-17", "categories": []}
{"id": "2305.10311", "title": "Investigating image-based fallow weed detection performance on Raphanus\n  sativus and Avena sativa at speeds up to 30 km h$^{-1}$", "abstract": "  Site-specific weed control (SSWC) can provide considerable reductions in weed\ncontrol costs and herbicide usage. Despite the promise of machine vision for\nSSWC systems and the importance of ground speed in weed control efficacy, there\nhas been little investigation of the role of ground speed and camera\ncharacteristics on weed detection performance. Here, we compare the performance\nof four camera-software combinations using the open-source OpenWeedLocator\nplatform - (1) default settings on a Raspberry Pi HQ camera, (2) optimised\nsoftware settings on a HQ camera, (3) optimised software settings on the\nRaspberry Pi v2 camera, and (4) a global shutter Arducam AR0234 camera - at\nspeeds ranging from 5 km h$^{-1}$ to 30 km h$^{-1}$. A combined excess green\n(ExG) and hue, saturation, value (HSV) thresholding algorithm was used for\ntesting under fallow conditions using tillage radish (Raphanus sativus) and\nforage oats (Avena sativa) as representative broadleaf and grass weeds,\nrespectively. ARD demonstrated the highest recall among camera systems, with up\nto 95.7% of weeds detected at 5 km h$^{-1}$ and 85.7% at 30 km h$^{-1}$. HQ1\nand V2 cameras had the lowest recall of 31.1% and 26.0% at 30 km h$^{-1}$,\nrespectively. All cameras experienced a decrease in recall as speed increased.\nThe highest rate of decrease was observed for HQ1 with 1.12% and 0.90%\nreductions in recall for every km h$^{-1}$ increase in speed for tillage radish\nand forage oats, respectively. Detection of the grassy forage oats was worse\n(P<0.05) than the broadleaved tillage radish for all cameras. Despite the\nvariations in recall, HQ1, HQ2, and V2 maintained near-perfect precision at all\ntested speeds. The variable effect of ground speed and camera system on\ndetection performance of grass and broadleaf weeds, indicates that careful\nhardware and software considerations must be made when developing SSWC systems.\n", "date": "2023-05-17", "categories": []}
{"id": "2303.02632", "title": "Deep-Learning-based Counting Methods, Datasets, and Applications in\n  Agriculture -- A Review", "abstract": "  The number of objects is considered an important factor in a variety of tasks\nin the agricultural domain. Automated counting can improve farmers decisions\nregarding yield estimation, stress detection, disease prevention, and more. In\nrecent years, deep learning has been increasingly applied to many\nagriculture-related applications, complementing conventional computer-vision\nalgorithms for counting agricultural objects. This article reviews progress in\nthe past decade and the state of the art for counting methods in agriculture,\nfocusing on deep-learning methods. It presents an overview of counting\nalgorithms, metrics, platforms, and sensors, a list of all publicly available\ndatasets, and an in-depth discussion of various deep-learning methods used for\ncounting. Finally, it discusses open challenges in object counting using deep\nlearning and gives a glimpse into new directions and future perspectives for\ncounting research. The review reveals a major leap forward in object counting\nin agriculture in the past decade, led by the penetration of deep learning\nmethods into counting platforms.\n", "date": "2023-03-05", "categories": []}
{"id": "2305.01823", "title": "Out-of-distribution detection algorithms for robust insect\n  classification", "abstract": "  Deep learning-based approaches have produced models with good insect\nclassification accuracy; Most of these models are conducive for application in\ncontrolled environmental conditions. One of the primary emphasis of researchers\nis to implement identification and classification models in the real\nagriculture fields, which is challenging because input images that are wildly\nout of the distribution (e.g., images like vehicles, animals, humans, or a\nblurred image of an insect or insect class that is not yet trained on) can\nproduce an incorrect insect classification. Out-of-distribution (OOD) detection\nalgorithms provide an exciting avenue to overcome these challenge as it ensures\nthat a model abstains from making incorrect classification prediction of\nnon-insect and/or untrained insect class images. We generate and evaluate the\nperformance of state-of-the-art OOD algorithms on insect detection classifiers.\nThese algorithms represent a diversity of methods for addressing an OOD\nproblem. Specifically, we focus on extrusive algorithms, i.e., algorithms that\nwrap around a well-trained classifier without the need for additional\nco-training. We compared three OOD detection algorithms: (i) Maximum Softmax\nProbability, which uses the softmax value as a confidence score, (ii)\nMahalanobis distance-based algorithm, which uses a generative classification\napproach; and (iii) Energy-Based algorithm that maps the input data to a scalar\nvalue, called energy. We performed an extensive series of evaluations of these\nOOD algorithms across three performance axes: (a) \\textit{Base model accuracy}:\nHow does the accuracy of the classifier impact OOD performance? (b) How does\nthe \\textit{level of dissimilarity to the domain} impact OOD performance? and\n(c) \\textit{Data imbalance}: How sensitive is OOD performance to the imbalance\nin per-class sample size?\n", "date": "2023-05-02", "categories": []}
{"id": "2304.09351", "title": "Machine Vision System for Early-stage Apple Flowers and Flower Clusters\n  Detection for Precision Thinning and Pollination", "abstract": "  Early-stage identification of fruit flowers that are in both opened and\nunopened condition in an orchard environment is significant information to\nperform crop load management operations such as flower thinning and pollination\nusing automated and robotic platforms. These operations are important in\ntree-fruit agriculture to enhance fruit quality, manage crop load, and enhance\nthe overall profit. The recent development in agricultural automation suggests\nthat this can be done using robotics which includes machine vision technology.\nIn this article, we proposed a vision system that detects early-stage flowers\nin an unstructured orchard environment using YOLOv5 object detection algorithm.\nFor the robotics implementation, the position of a cluster of the flower\nblossom is important to navigate the robot and the end effector. The centroid\nof individual flowers (both open and unopen) was identified and associated with\nflower clusters via K-means clustering. The accuracy of the opened and unopened\nflower detection is achieved up to mAP of 81.9% in commercial orchard images.\n", "date": "2023-04-19", "categories": []}
{"id": "2304.06177", "title": "Visual based Tomato Size Measurement System for an Indoor Farming\n  Environment", "abstract": "  As technology progresses, smart automated systems will serve an increasingly\nimportant role in the agricultural industry. Current existing vision systems\nfor yield estimation face difficulties in occlusion and scalability as they\nutilize a camera system that is large and expensive, which are unsuitable for\norchard environments. To overcome these problems, this paper presents a size\nmeasurement method combining a machine learning model and depth images captured\nfrom three low cost RGBD cameras to detect and measure the height and width of\ntomatoes. The performance of the presented system is evaluated on a lab\nenvironment with real tomato fruits and fake leaves to simulate occlusion in\nthe real farm environment. To improve accuracy by addressing fruit occlusion,\nour three-camera system was able to achieve a height measurement accuracy of\n0.9114 and a width accuracy of 0.9443.\n", "date": "2023-04-12", "categories": []}
{"id": "2304.04966", "title": "Computer Vision-Aided Intelligent Monitoring of Coffee: Towards\n  Sustainable Coffee Production", "abstract": "  Coffee which is prepared from the grinded roasted seeds of harvested coffee\ncherries, is one of the most consumed beverage and traded commodity, globally.\nTo manually monitor the coffee field regularly, and inform about plant and soil\nhealth, as well as estimate yield and harvesting time, is labor-intensive,\ntime-consuming and error-prone. Some recent studies have developed sensors for\nestimating coffee yield at the time of harvest, however a more inclusive and\napplicable technology to remotely monitor multiple parameters of the field and\nestimate coffee yield and quality even at pre-harvest stage, was missing.\nFollowing precision agriculture approach, we employed machine learning\nalgorithm YOLO, for image processing of coffee plant. In this study, the latest\nversion of the state-of-the-art algorithm YOLOv7 was trained with 324 annotated\nimages followed by its evaluation with 82 unannotated images as test data.\nNext, as an innovative approach for annotating the training data, we trained\nK-means models which led to machine-generated color classes of coffee fruit and\ncould thus characterize the informed objects in the image. Finally, we\nattempted to develop an AI-based handy mobile application which would not only\nefficiently predict harvest time, estimate coffee yield and quality, but also\ninform about plant health. Resultantly, the developed model efficiently\nanalyzed the test data with a mean average precision of 0.89. Strikingly, our\ninnovative semi-supervised method with an mean average precision of 0.77 for\nmulti-class mode surpassed the supervised method with mean average precision of\nonly 0.60, leading to faster and more accurate annotation. The mobile\napplication we designed based on the developed code, was named CoffeApp, which\npossesses multiple features of analyzing fruit from the image taken by phone\ncamera with in field and can thus track fruit ripening in real time.\n", "date": "2023-04-11", "categories": []}
{"id": "2304.03511", "title": "Carrot Cure: A CNN based Application to Detect Carrot Disease", "abstract": "  Carrot is a famous nutritional vegetable and developed all over the world.\nDifferent diseases of Carrot has become a massive issue in the carrot\nproduction circle which leads to a tremendous effect on the economic growth in\nthe agricultural sector. An automatic carrot disease detection system can help\nto identify malicious carrots and can provide a guide to cure carrot disease in\nan earlier stage, resulting in a less economical loss in the carrot production\nsystem. The proposed research study has developed a web application Carrot Cure\nbased on Convolutional Neural Network (CNN), which can identify a defective\ncarrot and provide a proper curative solution. Images of carrots affected by\ncavity spot and leaf bright as well as healthy images were collected. Further,\nthis research work has employed Convolutional Neural Network to include birth\nneural purposes and a Fully Convolutional Neural Network model (FCNN) for\ninfection order. Different avenues regarding different convolutional models\nwith colorful layers are explored and the proposed Convolutional model has\nachieved the perfection of 99.8%, which will be useful for the drovers to\ndistinguish carrot illness and boost their advantage.\n", "date": "2023-04-07", "categories": []}
{"id": "2304.03610", "title": "Look how they have grown: Non-destructive Leaf Detection and Size\n  Estimation of Tomato Plants for 3D Growth Monitoring", "abstract": "  Smart farming is a growing field as technology advances. Plant\ncharacteristics are crucial indicators for monitoring plant growth. Research\nhas been done to estimate characteristics like leaf area index, leaf disease,\nand plant height. However, few methods have been applied to non-destructive\nmeasurements of leaf size. In this paper, an automated non-destructive\nimaged-based measuring system is presented, which uses 2D and 3D data obtained\nusing a Zivid 3D camera, creating 3D virtual representations (digital twins) of\nthe tomato plants. Leaves are detected from corresponding 2D RGB images and\nmapped to their 3D point cloud using the detected leaf masks, which then pass\nthe leaf point cloud to the plane fitting algorithm to extract the leaf size to\nprovide data for growth monitoring. The performance of the measurement platform\nhas been measured through a comprehensive trial on real-world tomato plants\nwith quantified performance metrics compared to ground truth measurements.\nThree tomato leaf and height datasets (including 50+ 3D point cloud files of\ntomato plants) were collected and open-sourced in this project. The proposed\nleaf size estimation method demonstrates an RMSE value of 4.47mm and an R^2\nvalue of 0.87. The overall measurement system (leaf detection and size\nestimation algorithms combine) delivers an RMSE value of 8.13mm and an R^2\nvalue of 0.899.\n", "date": "2023-04-07", "categories": []}
{"id": "2304.02291", "title": "Trap-Based Pest Counting: Multiscale and Deformable Attention CenterNet\n  Integrating Internal LR and HR Joint Feature Learning", "abstract": "  Pest counting, which predicts the number of pests in the early stage, is very\nimportant because it enables rapid pest control, reduces damage to crops, and\nimproves productivity. In recent years, light traps have been increasingly used\nto lure and photograph pests for pest counting. However, pest images have a\nwide range of variability in pest appearance owing to severe occlusion, wide\npose variation, and even scale variation. This makes pest counting more\nchallenging. To address these issues, this study proposes a new pest counting\nmodel referred to as multiscale and deformable attention CenterNet\n(Mada-CenterNet) for internal low-resolution (LR) and high-resolution (HR)\njoint feature learning. Compared with the conventional CenterNet, the proposed\nMada-CenterNet adopts a multiscale heatmap generation approach in a two-step\nfashion to predict LR and HR heatmaps adaptively learned to scale variations,\nthat is, changes in the number of pests. In addition, to overcome the pose and\nocclusion problems, a new between-hourglass skip connection based on deformable\nand multiscale attention is designed to ensure internal LR and HR joint feature\nlearning and incorporate geometric deformation, thereby resulting in an\nimproved pest counting accuracy. Through experiments, the proposed\nMada-CenterNet is verified to generate the HR heatmap more accurately and\nimprove pest counting accuracy owing to multiscale heatmap generation, joint\ninternal feature learning, and deformable and multiscale attention. In\naddition, the proposed model is confirmed to be effective in overcoming severe\nocclusions and variations in pose and scale. The experimental results show that\nthe proposed model outperforms state-of-the-art crowd counting and object\ndetection models.\n", "date": "2023-04-05", "categories": []}
{"id": "2304.00763", "title": "BOLLWM: A real-world dataset for bollworm pest monitoring from cotton\n  fields in India", "abstract": "  This paper presents a dataset of agricultural pest images captured over five\nyears by thousands of small holder farmers and farming extension workers across\nIndia. The dataset has been used to support a mobile application that relies on\nartificial intelligence to assist farmers with pest management decisions.\nCreation came from a mix of organized data collection, and from mobile\napplication usage that was less controlled. This makes the dataset unique\nwithin the pest detection community, exhibiting a number of characteristics\nthat place it closer to other non-agricultural objected detection datasets.\nThis not only makes the dataset applicable to future pest management\napplications, it opens the door for a wide variety of other research agendas.\n", "date": "2023-04-03", "categories": []}
{"id": "2303.14347", "title": "Vision-based Vineyard Navigation Solution with Automatic Annotation", "abstract": "  Autonomous navigation is the key to achieving the full automation of\nagricultural research and production management (e.g., disease management and\nyield prediction) using agricultural robots. In this paper, we introduced a\nvision-based autonomous navigation framework for agriculture robots in\ntrellised cropping systems such as vineyards. To achieve this, we proposed a\nnovel learning-based method to estimate the path traversibility heatmap\ndirectly from an RGB-D image and subsequently convert the heatmap to a\npreferred traversal path. An automatic annotation pipeline was developed to\nform a training dataset by projecting RTK GPS paths collected during the first\nsetup in a vineyard in corresponding RGB-D images as ground-truth path\nannotations, allowing a fast model training and fine-tuning without costly\nhuman annotation. The trained path detection model was used to develop a full\nnavigation framework consisting of row tracking and row switching modules,\nenabling a robot to traverse within a crop row and transit between crop rows to\ncover an entire vineyard autonomously. Extensive field trials were conducted in\nthree different vineyards to demonstrate that the developed path detection\nmodel and navigation framework provided a cost-effective, accurate, and robust\nautonomous navigation solution in the vineyard and could be generalized to\nunseen vineyards with stable performance.\n", "date": "2023-03-25", "categories": []}
{"id": "2303.12499", "title": "On Domain-Specific Pre-Training for Effective Semantic Perception in\n  Agricultural Robotics", "abstract": "  Agricultural robots have the prospect to enable more efficient and\nsustainable agricultural production of food, feed, and fiber. Perception of\ncrops and weeds is a central component of agricultural robots that aim to\nmonitor fields and assess the plants as well as their growth stage in an\nautomatic manner. Semantic perception mostly relies on deep learning using\nsupervised approaches, which require time and qualified workers to label fairly\nlarge amounts of data. In this paper, we look into the problem of reducing the\namount of labels without compromising the final segmentation performance. For\nrobots operating in the field, pre-training networks in a supervised way is\nalready a popular method to reduce the number of required labeled images. We\ninvestigate the possibility of pre-training in a self-supervised fashion using\ndata from the target domain. To better exploit this data, we propose a set of\ndomain-specific augmentation strategies. We evaluate our pre-training on\nsemantic segmentation and leaf instance segmentation, two important tasks in\nour domain. The experimental results suggest that pre-training with\ndomain-specific data paired with our data augmentation strategy leads to\nsuperior performance compared to commonly used pre-trainings. Furthermore, the\npre-trained networks obtain similar performance to the fully supervised with\nless labeled data.\n", "date": "2023-03-22", "categories": []}
{"id": "2303.10444", "title": "Stall Number Detection of Cow Teats Key Frames", "abstract": "  In this paper, we present a small cow stall number dataset named\nCowStallNumbers, which is extracted from cow teat videos with the goal of\nadvancing cow stall number detection. This dataset contains 1042 training\nimages and 261 test images with the stall number ranging from 0 to 60. In\naddition, we fine-tuned a ResNet34 model and augmented the dataset with the\nrandom crop, center crop, and random rotation. The experimental result achieves\na 92% accuracy in stall number recognition and a 40.1% IoU score in stall\nnumber position prediction.\n", "date": "2023-03-18", "categories": []}
{"id": "2303.11527", "title": "Machine Learning Techniques for Estimating Soil Moisture from Mobile\n  Captured Images", "abstract": "  Precise Soil Moisture (SM) assessment is essential in agriculture. By\nunderstanding the level of SM, we can improve yield irrigation scheduling which\nsignificantly impacts food production and other needs of the global population.\nThe advancements in smartphone technologies and computer vision have\ndemonstrated a non-destructive nature of soil properties, including SM. The\nstudy aims to analyze the existing Machine Learning (ML) techniques for\nestimating SM from soil images and understand the moisture accuracy using\ndifferent smartphones and various sunlight conditions. Therefore, 629 images of\n38 soil samples were taken from seven areas in Sydney, Australia, and split\ninto four datasets based on the image-capturing devices used (iPhone 6s and\niPhone 11 Pro) and the lighting circumstances (direct and indirect sunlight). A\ncomparison between Multiple Linear Regression (MLR), Support Vector Regression\n(SVR), and Convolutional Neural Network (CNN) was presented. MLR was performed\nwith higher accuracy using holdout cross-validation, where the images were\ncaptured in indirect sunlight with the Mean Absolute Error (MAE) value of 0.35,\nRoot Mean Square Error (RMSE) value of 0.15, and R^2 value of 0.60.\nNevertheless, SVR was better with MAE, RMSE, and R^2 values of 0.05, 0.06, and\n0.96 for 10-fold cross-validation and 0.22, 0.06, and 0.95 for leave-one-out\ncross-validation when images were captured in indirect sunlight. It\ndemonstrates a smartphone camera's potential for predicting SM by utilizing ML.\nIn the future, software developers can develop mobile applications based on the\nresearch findings for accurate, easy, and rapid SM estimation.\n", "date": "2023-03-21", "categories": []}
{"id": "2303.10542", "title": "Wheat Head Counting by Estimating a Density Map with Convolutional\n  Neural Networks", "abstract": "  Wheat is one of the most significant crop species with an annual worldwide\ngrain production of 700 million tonnes. Assessing the production of wheat\nspikes can help us measure the grain production. Thus, detecting and\ncharacterizing spikes from images of wheat fields is an essential component in\na wheat breeding process. In this study, we propose three wheat head counting\nnetworks (WHCNet\\_1, WHCNet\\_2 and WHCNet\\_3) to accurately estimate the wheat\nhead count from an individual image and construct high quality density map,\nwhich illustrates the distribution of wheat heads in the image. The WHCNets are\ncomposed of two major components: a convolutional neural network (CNN) as the\nfront-end for wheat head image feature extraction and a CNN with skip\nconnections for the back-end to generate high-quality density maps. The dataset\nused in this study is the Global Wheat Head Detection (GWHD) dataset, which is\na large, diverse, and well-labelled dataset of wheat images and built by a\njoint international collaborative effort. We compare our methods with CSRNet, a\ndeep learning method which developed for highly congested scenes understanding\nand performing accurate count estimation as well as presenting high quality\ndensity maps. By taking the advantage of the skip connections between CNN\nlayers, WHCNets integrate features from low CNN layers to high CNN layers,\nthus, the output density maps have both high spatial resolution and detailed\nrepresentations of the input images. The experiments showed that our methods\noutperformed CSRNet in terms of the evaluation metrics, mean absolute error\n(MAE) and the root mean squared error (RMSE) with smaller model sizes. The code\nhas been deposited on GitHub (\\url{https://github.com/hyguozz}).\n", "date": "2023-03-19", "categories": []}
{"id": "2303.08415", "title": "Rice paddy disease classifications using CNNs", "abstract": "  Rice is a staple food in the world's diet, and yet huge percentages of crop\nyields are lost each year to disease. To combat this problem, people have been\nsearching for ways to automate disease diagnosis. Here, we extend on previous\nmodelling work by analysing how disease-classification accuracy is sensitive to\nboth model architecture and common computer vision techniques. In doing so, we\nmaximise accuracy whilst working in the constraints of smaller model sizes,\nminimum GPUs and shorter training times. Whilst previous state-of-the-art\nmodels had 93% accuracy only predicting 5 diseases, we improve this to 98.7%\nusing 10 disease classes.\n", "date": "2023-03-15", "categories": []}
{"id": "2211.14175", "title": "MCFFA-Net: Multi-Contextual Feature Fusion and Attention Guided Network\n  for Apple Foliar Disease Classification", "abstract": "  Numerous diseases cause severe economic loss in the apple production-based\nindustry. Early disease identification in apple leaves can help to stop the\nspread of infections and provide better productivity. Therefore, it is crucial\nto study the identification and classification of different apple foliar\ndiseases. Various traditional machine learning and deep learning methods have\naddressed and investigated this issue. However, it is still challenging to\nclassify these diseases because of their complex background, variation in the\ndiseased spot in the images, and the presence of several symptoms of multiple\ndiseases on the same leaf. This paper proposes a novel transfer learning-based\nstacked ensemble architecture named MCFFA-Net, which is composed of three\npre-trained architectures named MobileNetV2, DenseNet201, and InceptionResNetV2\nas backbone networks. We also propose a novel multi-scale dilated residual\nconvolution module to capture multi-scale contextual information with several\ndilated receptive fields from the extracted features. Channel-based attention\nmechanism is provided through squeeze and excitation networks to make the\nMCFFA-Net focused on the relevant information in the multi-receptive fields.\nThe proposed MCFFA-Net achieves a classification accuracy of 90.86%.\n", "date": "2022-11-25", "categories": []}
{"id": "2303.04884", "title": "O2RNet: Occluder-Occludee Relational Network for Robust Apple Detection\n  in Clustered Orchard Environments", "abstract": "  Automated apple harvesting has attracted significant research interest in\nrecent years due to its potential to revolutionize the apple industry,\naddressing the issues of shortage and high costs in labor. One key technology\nto fully enable efficient automated harvesting is accurate and robust apple\ndetection, which is challenging due to complex orchard environments that\ninvolve varying lighting conditions and foliage/branch occlusions. Furthermore,\nclustered apples are common in the orchard, which brings additional challenges\nas the clustered apples may be identified as one apple. This will cause issues\nin localization for subsequent robotic operations. In this paper, we present\nthe development of a novel deep learning-based apple detection framework,\nOccluder-Occludee Relational Network (O2RNet), for robust detection of apples\nin such clustered environments. This network exploits the occuluder-occludee\nrelationship modeling head by introducing a feature expansion structure to\nenable the combination of layered traditional detectors to split clustered\napples and foliage occlusions. More specifically, we collect a comprehensive\napple orchard image dataset under different lighting conditions (overcast,\nfront lighting, and back lighting) with frequent apple occlusions. We then\ndevelop a novel occlusion-aware network for apple detection, in which a feature\nexpansion structure is incorporated into the convolutional neural networks to\nextract additional features generated by the original network for occluded\napples. Comprehensive evaluations are performed, which show that the developed\nO2RNet outperforms state-of-the-art models with a higher accuracy of 94\\% and a\nhigher F1-score of 0.88 on apple detection.\n", "date": "2023-03-08", "categories": []}
{"id": "2212.14441", "title": "Fruit Ripeness Classification: a Survey", "abstract": "  Fruit is a key crop in worldwide agriculture feeding millions of people. The\nstandard supply chain of fruit products involves quality checks to guarantee\nfreshness, taste, and, most of all, safety. An important factor that determines\nfruit quality is its stage of ripening. This is usually manually classified by\nfield experts, making it a labor-intensive and error-prone process. Thus, there\nis an arising need for automation in fruit ripeness classification. Many\nautomatic methods have been proposed that employ a variety of feature\ndescriptors for the food item to be graded. Machine learning and deep learning\ntechniques dominate the top-performing methods. Furthermore, deep learning can\noperate on raw data and thus relieve the users from having to compute complex\nengineered features, which are often crop-specific. In this survey, we review\nthe latest methods proposed in the literature to automatize fruit ripeness\nclassification, highlighting the most common feature descriptors they operate\non.\n", "date": "2022-12-29", "categories": []}
{"id": "2303.00431", "title": "OliVaR: Improving Olive Variety Recognition using Deep Neural Networks", "abstract": "  The easy and accurate identification of varieties is fundamental in\nagriculture, especially in the olive sector, where more than 1200 olive\nvarieties are currently known worldwide. Varietal misidentification leads to\nmany potential problems for all the actors in the sector: farmers and nursery\nworkers may establish the wrong variety, leading to its maladaptation in the\nfield; olive oil and table olive producers may label and sell a non-authentic\nproduct; consumers may be misled; and breeders may commit errors during\ntargeted crossings between different varieties. To date, the standard for\nvarietal identification and certification consists of two methods:\nmorphological classification and genetic analysis. The morphological\nclassification consists of the visual pairwise comparison of different organs\nof the olive tree, where the most important organ is considered to be the\nendocarp. In contrast, different methods for genetic classification exist\n(RAPDs, SSR, and SNP). Both classification methods present advantages and\ndisadvantages. Visual morphological classification requires highly specialized\npersonnel and is prone to human error. Genetic identification methods are more\naccurate but incur a high cost and are difficult to implement. This paper\nintroduces OliVaR, a novel approach to olive varietal identification. OliVaR\nuses a teacher-student deep learning architecture to learn the defining\ncharacteristics of the endocarp of each specific olive variety and perform\nclassification. We construct what is, to the best of our knowledge, the largest\nolive variety dataset to date, comprising image data for 131 varieties from the\nMediterranean basin. We thoroughly test OliVaR on this dataset and show that it\ncorrectly predicts olive varieties with over 86% accuracy.\n", "date": "2023-03-01", "categories": []}
{"id": "2302.09716", "title": "Seeing the Fruit for the Leaves: Towards Automated Apple Fruitlet\n  Thinning", "abstract": "  Following a global trend, the lack of reliable access to skilled labour is\ncausing critical issues for the effective management of apple orchards. One of\nthe primary challenges is maintaining skilled human operators capable of making\nprecise fruitlet thinning decisions. Thinning requires accurately measuring the\ntrue crop load for individual apple trees to provide optimal thinning decisions\non an individual basis. A challenging task due to the dense foliage obscuring\nthe fruitlets within the tree structure. This paper presents the initial\ndesign, implementation, and evaluation details of the vision system for an\nautomatic apple fruitlet thinning robot to meet this need. The platform\nconsists of a UR5 robotic arm and stereo cameras which enable it to look around\nthe leaves to map the precise number and size of the fruitlets on the apple\nbranches. We show that this platform can measure the fruitlet load on the apple\ntree to with 84% accuracy in a real-world commercial apple orchard while being\n87% precise.\n", "date": "2023-02-20", "categories": []}
{"id": "2302.08818", "title": "Apple scab detection in orchards using deep learning on colour and\n  multispectral images", "abstract": "  Apple scab is a fungal disease caused by Venturia inaequalis. Disease is of\nparticular concern for growers, as it causes significant damage to fruit and\nleaves, leading to loss of fruit and yield. This article examines the ability\nof deep learning and hyperspectral imaging to accurately identify an apple\nsymptom infection in apple trees. In total, 168 image scenes were collected\nusing conventional RGB and Visible to Near-infrared (VIS-NIR) spectral imaging\n(8 channels) in infected orchards. Spectral data were preprocessed with an\nArtificial Neural Network (ANN) trained in segmentation to detect scab pixels\nbased on spectral information. Linear Discriminant Analysis (LDA) was used to\nfind the most discriminating channels in spectral data based on the healthy\nleaf and scab infested leaf spectra. Five combinations of false-colour images\nwere created from the spectral data and the segmentation net results. The\nimages were trained and evaluated with a modified version of the YOLOv5\nnetwork. Despite the promising results of deep learning using RGB images\n(P=0.8, mAP@50=0.73), the detection of apple scab in apple trees using\nmultispectral imaging proved to be a difficult task. The high-light environment\nof the open field made it difficult to collect a balanced spectrum from the\nmultispectral camera, since the infrared channel and the visible channels\nneeded to be constantly balanced so that they did not overexpose in the images.\n", "date": "2023-02-17", "categories": []}
{"id": "2302.06698", "title": "An Application of Deep Learning for Sweet Cherry Phenotyping using YOLO\n  Object Detection", "abstract": "  Tree fruit breeding is a long-term activity involving repeated measurements\nof various fruit quality traits on a large number of samples. These traits are\ntraditionally measured by manually counting the fruits, weighing to indirectly\nmeasure the fruit size, and fruit colour is classified subjectively into\ndifferent color categories using visual comparison to colour charts. These\nprocesses are slow, expensive and subject to evaluators' bias and fatigue.\nRecent advancements in deep learning can help automate this process. A method\nwas developed to automatically count the number of sweet cherry fruits in a\ncamera's field of view in real time using YOLOv3. A system capable of analyzing\nthe image data for other traits such as size and color was also developed using\nPython. The YOLO model obtained close to 99% accuracy in object detection and\ncounting of cherries and 90% on the Intersection over Union metric for object\nlocalization when extracting size and colour information. The model surpasses\nhuman performance and offers a significant improvement compared to manual\ncounting.\n", "date": "2023-02-13", "categories": []}
{"id": "2302.00164", "title": "Detection of Tomato Ripening Stages using Yolov3-tiny", "abstract": "  One of the most important agricultural products in Mexico is the tomato\n(Solanum lycopersicum), which occupies the 4th place national most produced\nproduct . Therefore, it is necessary to improve its production, building\nautomatic detection system that detect, classify an keep tacks of the fruits is\none way to archieve it. So, in this paper, we address the design of a computer\nvision system to detect tomatoes at different ripening stages. To solve the\nproblem, we use a neural network-based model for tomato classification and\ndetection. Specifically, we use the YOLOv3-tiny model because it is one of the\nlightest current deep neural networks. To train it, we perform two grid\nsearches testing several combinations of hyperparameters. Our experiments\nshowed an f1-score of 90.0% in the localization and classification of ripening\nstages in a custom dataset.\n", "date": "2023-02-01", "categories": []}
{"id": "2208.13001", "title": "Weakly and Semi-Supervised Detection, Segmentation and Tracking of Table\n  Grapes with Limited and Noisy Data", "abstract": "  Detection, segmentation and tracking of fruits and vegetables are three\nfundamental tasks for precision agriculture, enabling robotic harvesting and\nyield estimation applications. However, modern algorithms are data hungry and\nit is not always possible to gather enough data to apply the best performing\nsupervised approaches. Since data collection is an expensive and cumbersome\ntask, the enabling technologies for using computer vision in agriculture are\noften out of reach for small businesses. Following previous work in this\ncontext, where we proposed an initial weakly supervised solution to reduce the\ndata needed to get state-of-the-art detection and segmentation in precision\nagriculture applications, here we improve that system and explore the problem\nof tracking fruits in orchards. We present the case of vineyards of table\ngrapes in southern Lazio (Italy) since grapes are a difficult fruit to segment\ndue to occlusion, color and general illumination conditions. We consider the\ncase in which there is some initial labelled data that could work as source\ndata (\\eg wine grape data), but it is considerably different from the target\ndata (e.g. table grape data). To improve detection and segmentation on the\ntarget data, we propose to train the segmentation algorithm with a weak\nbounding box label, while for tracking we leverage 3D Structure from Motion\nalgorithms to generate new labels from already labelled samples. Finally, the\ntwo systems are combined in a full semi-supervised approach. Comparisons with\nstate-of-the-art supervised solutions show how our methods are able to train\nnew models that achieve high performances with few labelled images and with\nvery simple labelling.\n", "date": "2022-08-27", "categories": []}
{"id": "2301.08252", "title": "Evaluation of the potential of Near Infrared Hyperspectral Imaging for\n  monitoring the invasive brown marmorated stink bug", "abstract": "  The brown marmorated stink bug (BMSB), Halyomorpha halys, is an invasive\ninsect pest of global importance that damages several crops, compromising\nagri-food production. Field monitoring procedures are fundamental to perform\nrisk assessment operations, in order to promptly face crop infestations and\navoid economical losses. To improve pest management, spectral cameras mounted\non Unmanned Aerial Vehicles (UAVs) and other Internet of Things (IoT) devices,\nsuch as smart traps or unmanned ground vehicles, could be used as an innovative\ntechnology allowing fast, efficient and real-time monitoring of insect\ninfestations. The present study consists in a preliminary evaluation at the\nlaboratory level of Near Infrared Hyperspectral Imaging (NIR-HSI) as a possible\ntechnology to detect BMSB specimens on different vegetal backgrounds,\novercoming the problem of BMSB mimicry. Hyperspectral images of BMSB were\nacquired in the 980-1660 nm range, considering different vegetal backgrounds\nselected to mimic a real field application scene. Classification models were\nobtained following two different chemometric approaches. The first approach was\nfocused on modelling spectral information and selecting relevant spectral\nregions for discrimination by means of sparse-based variable selection coupled\nwith Soft Partial Least Squares Discriminant Analysis (s-Soft PLS-DA)\nclassification algorithm. The second approach was based on modelling spatial\nand spectral features contained in the hyperspectral images using Convolutional\nNeural Networks (CNN). Finally, to further improve BMSB detection ability, the\ntwo strategies were merged, considering only the spectral regions selected by\ns-Soft PLS-DA for CNN modelling.\n", "date": "2023-01-19", "categories": []}
{"id": "2207.04523", "title": "Facilitated machine learning for image-based fruit quality assessment", "abstract": "  Image-based machine learning models can be used to make the sorting and\ngrading of agricultural products more efficient. In many regions, implementing\nsuch systems can be difficult due to the lack of centralization and automation\nof postharvest supply chains. Stakeholders are often too small to specialize in\nmachine learning, and large training data sets are unavailable. We propose a\nmachine learning procedure for images based on pre-trained Vision Transformers.\nIt is easier to implement than the current standard approach of training\nConvolutional Neural Networks (CNNs) as we do not (re-)train deep neural\nnetworks. We evaluate our approach based on two data sets for apple defect\ndetection and banana ripeness estimation. Our model achieves a competitive\nclassification accuracy equal to or less than one percent below the\nbest-performing CNN. At the same time, it requires three times fewer training\nsamples to achieve a 90% accuracy.\n", "date": "2022-07-10", "categories": []}
{"id": "2301.03947", "title": "Autonomous Strawberry Picking Robotic System (Robofruit)", "abstract": "  Challenges in strawberry picking made selective harvesting robotic technology\ndemanding. However, selective harvesting of strawberries is complicated forming\na few scientific research questions. Most available solutions only deal with a\nspecific picking scenario, e.g., picking only a single variety of fruit in\nisolation. Nonetheless, most economically viable (e.g. high-yielding and/or\ndisease-resistant) varieties of strawberry are grown in dense clusters. The\ncurrent perception technology in such use cases is inefficient. In this work,\nwe developed a novel system capable of harvesting strawberries with several\nunique features. The features allow the system to deal with very complex\npicking scenarios, e.g. dense clusters. Our concept of a modular system makes\nour system reconfigurable to adapt to different picking scenarios. We designed,\nmanufactured, and tested a picking head with 2.5 DOF (2 independent mechanisms\nand 1 dependent cutting system) capable of removing possible occlusions and\nharvesting targeted strawberries without contacting fruit flesh to avoid damage\nand bruising. In addition, we developed a novel perception system to localise\nstrawberries and detect their key points, picking points, and determine their\nripeness. For this purpose, we introduced two new datasets. Finally, we tested\nthe system in a commercial strawberry growing field and our research farm with\nthree different strawberry varieties. The results show the effectiveness and\nreliability of the proposed system. The designed picking head was able to\nremove occlusions and harvest strawberries effectively. The perception system\nwas able to detect and determine the ripeness of strawberries with 95%\naccuracy. In total, the system was able to harvest 87% of all detected\nstrawberries with a success rate of 83% for all pluckable fruits. We also\ndiscuss a series of open research questions in the discussion section.\n", "date": "2023-01-10", "categories": []}
{"id": "2212.04255", "title": "Fruit Quality Assessment with Densely Connected Convolutional Neural\n  Network", "abstract": "  Accurate recognition of food items along with quality assessment is of\nparamount importance in the agricultural industry. Such automated systems can\nspeed up the wheel of the food processing sector and save tons of manual labor.\nIn this connection, the recent advancement of Deep learning-based architectures\nhas introduced a wide variety of solutions offering remarkable performance in\nseveral classification tasks. In this work, we have exploited the concept of\nDensely Connected Convolutional Neural Networks (DenseNets) for fruit quality\nassessment. The feature propagation towards the deeper layers has enabled the\nnetwork to tackle the vanishing gradient problems and ensured the reuse of\nfeatures to learn meaningful insights. Evaluating on a dataset of 19,526 images\ncontaining six fruits having three quality grades for each, the proposed\npipeline achieved a remarkable accuracy of 99.67%. The robustness of the model\nwas further tested for fruit classification and quality assessment tasks where\nthe model produced a similar performance, which makes it suitable for real-life\napplications.\n", "date": "2022-12-08", "categories": []}
{"id": "2212.12856", "title": "Deep Cost-sensitive Learning for Wheat Frost Detection", "abstract": "  Frost damage is one of the main factors leading to wheat yield reduction.\nTherefore, the detection of wheat frost accurately and efficiently is\nbeneficial for growers to take corresponding measures in time to reduce\neconomic loss. To detect the wheat frost, in this paper we create a\nhyperspectral wheat frost data set by collecting the data characterized by\ntemperature, wheat yield, and hyperspectral information provided by the\nhandheld hyperspectral spectrometer. However, due to the imbalance of data,\nthat is, the number of healthy samples is much higher than the number of frost\ndamage samples, a deep learning algorithm tends to predict biasedly towards the\nhealthy samples resulting in model overfitting of the healthy samples.\nTherefore, we propose a method based on deep cost-sensitive learning, which\nuses a one-dimensional convolutional neural network as the basic framework and\nincorporates cost-sensitive learning with fixed factors and adjustment factors\ninto the loss function to train the network. Meanwhile, the accuracy and score\nare used as evaluation metrics. Experimental results show that the detection\naccuracy and the score reached 0.943 and 0.623 respectively, this demonstration\nshows that this method not only ensures the overall accuracy but also\neffectively improves the detection rate of frost samples.\n", "date": "2022-12-25", "categories": []}
{"id": "2212.07527", "title": "Plastic Contaminant Detection in Aerial Imagery of Cotton Fields with\n  Deep Learning", "abstract": "  Plastic shopping bags that get carried away from the side of roads and\ntangled on cotton plants can end up at cotton gins if not removed before the\nharvest. Such bags may not only cause problem in the ginning process but might\nalso get embodied in cotton fibers reducing its quality and marketable value.\nTherefore, it is required to detect, locate, and remove the bags before cotton\nis harvested. Manually detecting and locating these bags in cotton fields is\nlabor intensive, time-consuming and a costly process. To solve these\nchallenges, we present application of four variants of YOLOv5 (YOLOv5s,\nYOLOv5m, YOLOv5l and YOLOv5x) for detecting plastic shopping bags using\nUnmanned Aircraft Systems (UAS)-acquired RGB (Red, Green, and Blue) images. We\nalso show fixed effect model tests of color of plastic bags as well as\nYOLOv5-variant on average precision (AP), mean average precision (mAP@50) and\naccuracy. In addition, we also demonstrate the effect of height of plastic bags\non the detection accuracy. It was found that color of bags had significant\neffect (p < 0.001) on accuracy across all the four variants while it did not\nshow any significant effect on the AP with YOLOv5m (p = 0.10) and YOLOv5x (p =\n0.35) at 95% confidence level. Similarly, YOLOv5-variant did not show any\nsignificant effect on the AP (p = 0.11) and accuracy (p = 0.73) of white bags,\nbut it had significant effects on the AP (p = 0.03) and accuracy (p = 0.02) of\nbrown bags including on the mAP@50 (p = 0.01) and inference speed (p < 0.0001).\nAdditionally, height of plastic bags had significant effect (p < 0.0001) on\noverall detection accuracy. The findings reported in this paper can be useful\nin speeding up removal of plastic bags from cotton fields before harvest and\nthereby reducing the amount of contaminants that end up at cotton gins.\n", "date": "2022-12-14", "categories": []}
{"id": "2211.08743", "title": "Yield Evaluation of Citrus Fruits based on the YoloV5 compressed by\n  Knowledge Distillation", "abstract": "  In the field of planting fruit trees, pre-harvest estimation of fruit yield\nis important for fruit storage and price evaluation. However, considering the\ncost, the yield of each tree cannot be assessed by directly picking the\nimmature fruit. Therefore, the problem is a very difficult task. In this paper,\na fruit counting and yield assessment method based on computer vision is\nproposed for citrus fruit trees as an example. Firstly, images of single fruit\ntrees from different angles are acquired and the number of fruits is detected\nusing a deep Convolutional Neural Network model YOLOv5, and the model is\ncompressed using a knowledge distillation method. Then, a linear regression\nmethod is used to model yield-related features and evaluate yield. Experiments\nshow that the proposed method can accurately count fruits and approximate the\nyield.\n", "date": "2022-11-16", "categories": []}
{"id": "2211.04755", "title": "Towards Global Crop Maps with Transfer Learning", "abstract": "  The continuous increase in global population and the impact of climate change\non crop production are expected to affect the food sector significantly. In\nthis context, there is need for timely, large-scale and precise mapping of\ncrops for evidence-based decision making. A key enabler towards this direction\nare new satellite missions that freely offer big remote sensing data of high\nspatio-temporal resolution and global coverage. During the previous decade and\nbecause of this surge of big Earth observations, deep learning methods have\ndominated the remote sensing and crop mapping literature. Nevertheless, deep\nlearning models require large amounts of annotated data that are scarce and\nhard-to-acquire. To address this problem, transfer learning methods can be used\nto exploit available annotations and enable crop mapping for other regions,\ncrop types and years of inspection. In this work, we have developed and trained\na deep learning model for paddy rice detection in South Korea using Sentinel-1\nVH time-series. We then fine-tune the model for i) paddy rice detection in\nFrance and Spain and ii) barley detection in the Netherlands. Additionally, we\npropose a modification in the pre-trained weights in order to incorporate extra\ninput features (Sentinel-1 VV). Our approach shows excellent performance when\ntransferring in different areas for the same crop type and rather promising\nresults when transferring in a different area and crop type.\n", "date": "2022-11-09", "categories": []}
{"id": "2210.16923", "title": "See as a Bee: UV Sensor for Aerial Strawberry Crop Monitoring", "abstract": "  Precision agriculture aims to use technological tools for the agro-food\nsector to increase productivity, cut labor costs, and reduce the use of\nresources. This work takes inspiration from bees vision to design a remote\nsensing system tailored to incorporate UV-reflectance into a flower detector.\nWe demonstrate how this approach can provide feature-rich images for deep\nlearning strawberry flower detection and we apply it to a scalable, yet cost\neffective aerial monitoring robotic system in the field. We also compare the\nperformance of our UV-G-B image detector with a similar work that utilizes RGB\nimages.\n", "date": "2022-10-30", "categories": []}
{"id": "2210.13296", "title": "Semantic Image Segmentation with Deep Learning for Vine Leaf Phenotyping", "abstract": "  Plant phenotyping refers to a quantitative description of the plants\nproperties, however in image-based phenotyping analysis, our focus is primarily\non the plants anatomical, ontogenetical and physiological properties.This\ntechnique reinforced by the success of Deep Learning in the field of image\nbased analysis is applicable to a wide range of research areas making\nhigh-throughput screens of plants possible, reducing the time and effort needed\nfor phenotypic characterization.In this study, we use Deep Learning methods\n(supervised and unsupervised learning based approaches) to semantically segment\ngrapevine leaves images in order to develop an automated object detection\n(through segmentation) system for leaf phenotyping which will yield information\nregarding their structure and function.In these directions we studied several\ndeep learning approaches with promising results as well as we reported some\nfuture challenging tasks in the area of precision agriculture.Our work\ncontributes to plant lifecycle monitoring through which dynamic traits such as\ngrowth and development can be captured and quantified, targeted intervention\nand selective application of agrochemicals and grapevine variety identification\nwhich are key prerequisites in sustainable agriculture.\n", "date": "2022-10-24", "categories": []}
{"id": "2210.11545", "title": "Transferring learned patterns from ground-based field imagery to predict\n  UAV-based imagery for crop and weed semantic segmentation in precision crop\n  farming", "abstract": "  Weed and crop segmentation is becoming an increasingly integral part of\nprecision farming that leverages the current computer vision and deep learning\ntechnologies. Research has been extensively carried out based on images\ncaptured with a camera from various platforms. Unmanned aerial vehicles (UAVs)\nand ground-based vehicles including agricultural robots are the two popular\nplatforms for data collection in fields. They all contribute to site-specific\nweed management (SSWM) to maintain crop yield. Currently, the data from these\ntwo platforms is processed separately, though sharing the same semantic objects\n(weed and crop). In our paper, we have developed a deep convolutional network\nthat enables to predict both field and aerial images from UAVs for weed\nsegmentation and mapping with only field images provided in the training phase.\nThe network learning process is visualized by feature maps at shallow and deep\nlayers. The results show that the mean intersection of union (IOU) values of\nthe segmentation for the crop (maize), weeds, and soil background in the\ndeveloped model for the field dataset are 0.744, 0.577, 0.979, respectively,\nand the performance of aerial images from an UAV with the same model, the IOU\nvalues of the segmentation for the crop (maize), weeds and soil background are\n0.596, 0.407, and 0.875, respectively. To estimate the effect on the use of\nplant protection agents, we quantify the relationship between herbicide\nspraying saving rate and grid size (spraying resolution) based on the predicted\nweed map. The spraying saving rate is up to 90% when the spraying resolution is\nat 1.78 x 1.78 cm2. The study shows that the developed deep convolutional\nneural network could be used to classify weeds from both field and aerial\nimages and delivers satisfactory results.\n", "date": "2022-10-20", "categories": []}
{"id": "2210.09509", "title": "Deep Data Augmentation for Weed Recognition Enhancement: A Diffusion\n  Probabilistic Model and Transfer Learning Based Approach", "abstract": "  Weed management plays an important role in many modern agricultural\napplications. Conventional weed control methods mainly rely on chemical\nherbicides or hand weeding, which are often cost-ineffective, environmentally\nunfriendly, or even posing a threat to food safety and human health. Recently,\nautomated/robotic weeding using machine vision systems has seen increased\nresearch attention with its potential for precise and individualized weed\ntreatment. However, dedicated, large-scale, and labeled weed image datasets are\nrequired to develop robust and effective weed identification systems but they\nare often difficult and expensive to obtain. To address this issue, data\naugmentation approaches, such as generative adversarial networks (GANs), have\nbeen explored to generate highly realistic images for agricultural\napplications. Yet, despite some progress, those approaches are often\ncomplicated to train or have difficulties preserving fine details in images. In\nthis paper, we present the first work of applying diffusion probabilistic\nmodels (also known as diffusion models) to generate high-quality synthetic weed\nimages based on transfer learning. Comprehensive experimental results show that\nthe developed approach consistently outperforms several state-of-the-art GAN\nmodels, representing the best trade-off between sample fidelity and diversity\nand highest FID score on a common weed dataset, CottonWeedID15. In addition,\nthe expanding dataset with synthetic weed images can apparently boost model\nperformance on four deep learning (DL) models for the weed classification\ntasks. Furthermore, the DL models trained on CottonWeedID15 dataset with only\n10% of real images and 90% of synthetic weed images achieve a testing accuracy\nof over 94%, showing high-quality of the generated weed samples. The codes of\nthis study are made publicly available at\nhttps://github.com/DongChen06/DMWeeds.\n", "date": "2022-10-18", "categories": []}
{"id": "2207.03881", "title": "The Power of Transfer Learning in Agricultural Applications: AgriNet", "abstract": "  Advances in deep learning and transfer learning have paved the way for\nvarious automation classification tasks in agriculture, including plant\ndiseases, pests, weeds, and plant species detection. However, agriculture\nautomation still faces various challenges, such as the limited size of datasets\nand the absence of plant-domain-specific pretrained models. Domain specific\npretrained models have shown state of art performance in various computer\nvision tasks including face recognition and medical imaging diagnosis. In this\npaper, we propose AgriNet dataset, a collection of 160k agricultural images\nfrom more than 19 geographical locations, several images captioning devices,\nand more than 423 classes of plant species and diseases. We also introduce\nAgriNet models, a set of pretrained models on five ImageNet architectures:\nVGG16, VGG19, Inception-v3, InceptionResNet-v2, and Xception. AgriNet-VGG19\nachieved the highest classification accuracy of 94 % and the highest F1-score\nof 92%. Additionally, all proposed models were found to accurately classify the\n423 classes of plant species, diseases, pests, and weeds with a minimum\naccuracy of 87% for the Inception-v3 model.Finally, experiments to evaluate of\nsuperiority of AgriNet models compared to ImageNet models were conducted on two\nexternal datasets: pest and plant diseases dataset from Bangladesh and a plant\ndiseases dataset from Kashmir.\n", "date": "2022-07-08", "categories": []}
{"id": "2210.02665", "title": "Vision-Based Defect Classification and Weight Estimation of Rice Kernels", "abstract": "  Rice is one of the main staple food in many areas of the world. The quality\nestimation of rice kernels are crucial in terms of both food safety and\nsocio-economic impact. This was usually carried out by quality inspectors in\nthe past, which may result in both objective and subjective inaccuracies. In\nthis paper, we present an automatic visual quality estimation system of rice\nkernels, to classify the sampled rice kernels according to their types of\nflaws, and evaluate their quality via the weight ratios of the perspective\nkernel types. To compensate for the imbalance of different kernel numbers and\nclassify kernels with multiple flaws accurately, we propose a multi-stage\nworkflow which is able to locate the kernels in the captured image and classify\ntheir properties. We define a novel metric to measure the relative weight of\neach kernel in the image from its area, such that the relative weight of each\ntype of kernels with regard to the all samples can be computed and used as the\nbasis for rice quality estimation. Various experiments are carried out to show\nthat our system is able to output precise results in a contactless way and\nreplace tedious and error-prone manual works.\n", "date": "2022-10-06", "categories": []}
{"id": "2209.07683", "title": "SQ-Swin: a Pretrained Siamese Quadratic Swin Transformer for Lettuce\n  Browning Prediction", "abstract": "  Packaged fresh-cut lettuce is widely consumed as a major component of\nvegetable salad owing to its high nutrition, freshness, and convenience.\nHowever, enzymatic browning discoloration on lettuce cut edges significantly\nreduces product quality and shelf life. While there are many research and\nbreeding efforts underway to minimize browning, the progress is hindered by the\nlack of a rapid and reliable methodology to evaluate browning. Current methods\nto identify and quantify browning are either too subjective, labor intensive,\nor inaccurate. In this paper, we report a deep learning model for lettuce\nbrowning prediction. To the best of our knowledge, it is the first-of-its-kind\non deep learning for lettuce browning prediction using a pretrained Siamese\nQuadratic Swin (SQ-Swin) transformer with several highlights. First, our model\nincludes quadratic features in the transformer model which is more powerful to\nincorporate real-world representations than the linear transformer. Second, a\nmulti-scale training strategy is proposed to augment the data and explore more\nof the inherent self-similarity of the lettuce images. Third, the proposed\nmodel uses a siamese architecture which learns the inter-relations among the\nlimited training samples. Fourth, the model is pretrained on the ImageNet and\nthen trained with the reptile meta-learning algorithm to learn higher-order\ngradients than a regular one. Experiment results on the fresh-cut lettuce\ndatasets show that the proposed SQ-Swin outperforms the traditional methods and\nother deep learning-based backbones.\n", "date": "2022-09-16", "categories": []}
{"id": "2209.05282", "title": "Leveraging Artificial Intelligence Techniques for Smart Palm Tree\n  Detection: A Decade Systematic Review", "abstract": "  Over the past few years, total financial investment in the agricultural\nsector has increased substantially. Palm tree is important for many countries'\neconomies, particularly in northern Africa and the Middle East. Monitoring in\nterms of detection and counting palm trees provides useful information for\nvarious stakeholders; it helps in yield estimation and examination to ensure\nbetter crop quality and prevent pests, diseases, better irrigation, and other\npotential threats. Despite their importance, this information is still\nchallenging to obtain. This study systematically reviews research articles\nbetween 2011 and 2021 on artificial intelligence (AI) technology for smart palm\ntree detection. A systematic review (SR) was performed using the PRISMA\napproach based on a four-stage selection process. Twenty-two articles were\nincluded for the synthesis activity reached from the search strategy alongside\nthe inclusion criteria in order to answer to two main research questions. The\nstudy's findings reveal patterns, relationships, networks, and trends in\napplying artificial intelligence in palm tree detection over the last decade.\nDespite the good results in most of the studies, the effective and efficient\nmanagement of large-scale palm plantations is still a challenge. In addition,\ncountries whose economies strongly related to intelligent palm services,\nespecially in North Africa, should give more attention to this kind of study.\nThe results of this research could benefit both the research community and\nstakeholders.\n", "date": "2022-09-12", "categories": []}
{"id": "2209.02377", "title": "MangoLeafBD: A Comprehensive Image Dataset to Classify Diseased and\n  Healthy Mango Leaves", "abstract": "  Agriculture is of one of the few remaining sectors that is yet to receive\nproper attention from the machine learning community. The importance of\ndatasets in the machine learning discipline cannot be overemphasized. The lack\nof standard and publicly available datasets related to agriculture impedes\npractitioners of this discipline to harness the full benefit of these powerful\ncomputational predictive tools and techniques. To improve this scenario, we\ndevelop, to the best of our knowledge, the first-ever standard, ready-to-use,\nand publicly available dataset of mango leaves. The images are collected from\nfour mango orchards of Bangladesh, one of the top mango-growing countries of\nthe world. The dataset contains 4000 images of about 1800 distinct leaves\ncovering seven diseases. Although the dataset is developed using mango leaves\nof Bangladesh only, since we deal with diseases that are common across many\ncountries, this dataset is likely to be applicable to identify mango diseases\nin other countries as well, thereby boosting mango yield. This dataset is\nexpected to draw wide attention from machine learning researchers and\npractitioners in the field of automated agriculture.\n", "date": "2022-08-27", "categories": []}
{"id": "2208.12613", "title": "Image augmentation improves few-shot classification performance in plant\n  disease recognition", "abstract": "  With the world population projected to near 10 billion by 2050, minimizing\ncrop damage and guaranteeing food security has never been more important.\nMachine learning has been proposed as a solution to quickly and efficiently\nidentify diseases in crops. Convolutional Neural Networks typically require\nlarge datasets of annotated data which are not available on demand. Collecting\nthis data is a long and arduous process which involves manually picking,\nimaging, and annotating each individual leaf. I tackle the problem of plant\nimage data scarcity by exploring the efficacy of various data augmentation\ntechniques when used in conjunction with transfer learning. I evaluate the\nimpact of various data augmentation techniques both individually and combined\non the performance of a ResNet. I propose an augmentation scheme utilizing a\nsequence of different augmentations which consistently improves accuracy\nthrough many trials. Using only 10 total seed images, I demonstrate that my\naugmentation framework can increase model accuracy by upwards of 25\\%.\n", "date": "2022-08-25", "categories": []}
{"id": "2208.11566", "title": "Apple Counting using Convolutional Neural Networks", "abstract": "  Estimating accurate and reliable fruit and vegetable counts from images in\nreal-world settings, such as orchards, is a challenging problem that has\nreceived significant recent attention. Estimating fruit counts before harvest\nprovides useful information for logistics planning. While considerable progress\nhas been made toward fruit detection, estimating the actual counts remains\nchallenging. In practice, fruits are often clustered together. Therefore,\nmethods that only detect fruits fail to offer general solutions to estimate\naccurate fruit counts. Furthermore, in horticultural studies, rather than a\nsingle yield estimate, finer information such as the distribution of the number\nof apples per cluster is desirable. In this work, we formulate fruit counting\nfrom images as a multi-class classification problem and solve it by training a\nConvolutional Neural Network. We first evaluate the per-image accuracy of our\nmethod and compare it with a state-of-the-art method based on Gaussian Mixture\nModels over four test datasets. Even though the parameters of the Gaussian\nMixture Model-based method are specifically tuned for each dataset, our network\noutperforms it in three out of four datasets with a maximum of 94\\% accuracy.\nNext, we use the method to estimate the yield for two datasets for which we\nhave ground truth. Our method achieved 96-97\\% accuracies. For additional\ndetails please see our video here:\nhttps://www.youtube.com/watch?v=Le0mb5P-SYc}{https://www.youtube.com/watch?v=Le0mb5P-SYc.\n", "date": "2022-08-24", "categories": []}
{"id": "2208.10472", "title": "Automated Pruning of Polyculture Plants", "abstract": "  Polyculture farming has environmental advantages but requires substantially\nmore pruning than monoculture farming. We present novel hardware and algorithms\nfor automated pruning. Using an overhead camera to collect data from a physical\nscale garden testbed, the autonomous system utilizes a learned Plant\nPhenotyping convolutional neural network and a Bounding Disk Tracking algorithm\nto evaluate the individual plant distribution and estimate the state of the\ngarden each day. From this garden state, AlphaGardenSim selects plants to\nautonomously prune. A trained neural network detects and targets specific prune\npoints on the plant. Two custom-designed pruning tools, compatible with a\nFarmBot gantry system, are experimentally evaluated and execute autonomous cuts\nthrough controlled algorithms. We present results for four 60-day garden\ncycles. Results suggest the system can autonomously achieve 0.94 normalized\nplant diversity with pruning shears while maintaining an average canopy\ncoverage of 0.84 by the end of the cycles. For code, videos, and datasets, see\nhttps://sites.google.com/berkeley.edu/pruningpolyculture.\n", "date": "2022-08-22", "categories": []}
{"id": "2208.02394", "title": "End-to-end deep learning for directly estimating grape yield from\n  ground-based imagery", "abstract": "  Yield estimation is a powerful tool in vineyard management, as it allows\ngrowers to fine-tune practices to optimize yield and quality. However, yield\nestimation is currently performed using manual sampling, which is\ntime-consuming and imprecise. This study demonstrates the application of\nproximal imaging combined with deep learning for yield estimation in vineyards.\nContinuous data collection using a vehicle-mounted sensing kit combined with\ncollection of ground truth yield data at harvest using a commercial yield\nmonitor allowed for the generation of a large dataset of 23,581 yield points\nand 107,933 images. Moreover, this study was conducted in a mechanically\nmanaged commercial vineyard, representing a challenging environment for image\nanalysis but a common set of conditions in the California Central Valley. Three\nmodel architectures were tested: object detection, CNN regression, and\ntransformer models. The object detection model was trained on hand-labeled\nimages to localize grape bunches, and either bunch count or pixel area was\nsummed to correlate with grape yield. Conversely, regression models were\ntrained end-to-end to predict grape yield from image data without the need for\nhand labeling. Results demonstrated that both a transformer as well as the\nobject detection model with pixel area processing performed comparably, with a\nmean absolute percent error of 18% and 18.5%, respectively on a representative\nholdout dataset. Saliency mapping was used to demonstrate the attention of the\nCNN model was localized near the predicted location of grape bunches, as well\nas on the top of the grapevine canopy. Overall, the study showed the\napplicability of proximal imaging and deep learning for prediction of grapevine\nyield on a large scale. Additionally, the end-to-end modeling approach was able\nto perform comparably to the object detection approach while eliminating the\nneed for hand-labeling.\n", "date": "2022-08-04", "categories": []}
{"id": "2208.02707", "title": "Standardizing and Centralizing Datasets to Enable Efficient Training of\n  Agricultural Deep Learning Models", "abstract": "  In recent years, deep learning models have become the standard for\nagricultural computer vision. Such models are typically fine-tuned to\nagricultural tasks using model weights that were originally fit to more\ngeneral, non-agricultural datasets. This lack of agriculture-specific\nfine-tuning potentially increases training time and resource use, and decreases\nmodel performance, leading an overall decrease in data efficiency. To overcome\nthis limitation, we collect a wide range of existing public datasets for three\ndistinct tasks, standardize them, and construct standard training and\nevaluation pipelines, providing us with a set of benchmarks and pretrained\nmodels. We then conduct a number of experiments using methods which are\ncommonly used in deep learning tasks, but unexplored in their domain-specific\napplications for agriculture. Our experiments guide us in developing a number\nof approaches to improve data efficiency when training agricultural deep\nlearning models, without large-scale modifications to existing pipelines. Our\nresults demonstrate that even slight training modifications, such as using\nagricultural pretrained model weights, or adopting specific spatial\naugmentations into data processing pipelines, can significantly boost model\nperformance and result in shorter convergence time, saving training resources.\nFurthermore, we find that even models trained on low-quality annotations can\nproduce comparable levels of performance to their high-quality equivalents,\nsuggesting that datasets with poor annotations can still be used for training,\nexpanding the pool of currently available datasets. Our methods are broadly\napplicable throughout agricultural deep learning, and present high potential\nfor significant data efficiency improvements.\n", "date": "2022-08-04", "categories": []}
{"id": "2208.00002", "title": "HOB-CNN: Hallucination of Occluded Branches with a Convolutional Neural\n  Network for 2D Fruit Trees", "abstract": "  Orchard automation has attracted the attention of researchers recently due to\nthe shortage of global labor force. To automate tasks in orchards such as\npruning, thinning, and harvesting, a detailed understanding of the tree\nstructure is required. However, occlusions from foliage and fruits can make it\nchallenging to predict the position of occluded trunks and branches. This work\nproposes a regression-based deep learning model, Hallucination of Occluded\nBranch Convolutional Neural Network (HOB-CNN), for tree branch position\nprediction in varying occluded conditions. We formulate tree branch position\nprediction as a regression problem towards the horizontal locations of the\nbranch along the vertical direction or vice versa. We present comparative\nexperiments on Y-shaped trees with two state-of-the-art baselines, representing\ncommon approaches to the problem. Experiments show that HOB-CNN outperform the\nbaselines at predicting branch position and shows robustness against varying\nlevels of occlusion. We further validated HOB-CNN against two different types\nof 2D trees, and HOB-CNN shows generalization across different trees and\nrobustness under different occluded conditions.\n", "date": "2022-07-28", "categories": []}
{"id": "2208.00519", "title": "Assessing The Performance of YOLOv5 Algorithm for Detecting Volunteer\n  Cotton Plants in Corn Fields at Three Different Growth Stages", "abstract": "  The boll weevil (Anthonomus grandis L.) is a serious pest that primarily\nfeeds on cotton plants. In places like Lower Rio Grande Valley of Texas, due to\nsub-tropical climatic conditions, cotton plants can grow year-round and\ntherefore the left-over seeds from the previous season during harvest can\ncontinue to grow in the middle of rotation crops like corn (Zea mays L.) and\nsorghum (Sorghum bicolor L.). These feral or volunteer cotton (VC) plants when\nreach the pinhead squaring phase (5-6 leaf stage) can act as hosts for the boll\nweevil pest. The Texas Boll Weevil Eradication Program (TBWEP) employs people\nto locate and eliminate VC plants growing by the side of roads or fields with\nrotation crops but the ones growing in the middle of fields remain undetected.\nIn this paper, we demonstrate the application of computer vision (CV) algorithm\nbased on You Only Look Once version 5 (YOLOv5) for detecting VC plants growing\nin the middle of corn fields at three different growth stages (V3, V6, and VT)\nusing unmanned aircraft systems (UAS) remote sensing imagery. All the four\nvariants of YOLOv5 (s, m, l, and x) were used and their performances were\ncompared based on classification accuracy, mean average precision (mAP), and\nF1-score. It was found that YOLOv5s could detect VC plants with a maximum\nclassification accuracy of 98% and mAP of 96.3 % at the V6 stage of corn while\nYOLOv5s and YOLOv5m resulted in the lowest classification accuracy of 85% and\nYOLOv5m and YOLOv5l had the least mAP of 86.5% at the VT stage on images of\nsize 416 x 416 pixels. The developed CV algorithm has the potential to\neffectively detect and locate VC plants growing in the middle of corn fields as\nwell as expedite the management aspects of TBWEP.\n", "date": "2022-07-31", "categories": []}
{"id": "2207.02721", "title": "Deep Learning approach for Classifying Trusses and Runners of\n  Strawberries", "abstract": "  The use of artificial intelligence in the agricultural sector has been\ngrowing at a rapid rate to automate farming activities. Emergent farming\ntechnologies focus on mapping and classification of plants, fruits, diseases,\nand soil types. Although, assisted harvesting and pruning applications using\ndeep learning algorithms are in the early development stages, there is a demand\nfor solutions to automate such processes. This paper proposes the use of Deep\nLearning for the classification of trusses and runners of strawberry plants\nusing semantic segmentation and dataset augmentation. The proposed approach is\nbased on the use of noises (i.e. Gaussian, Speckle, Poisson and\nSalt-and-Pepper) to artificially augment the dataset and compensate the low\nnumber of data samples and increase the overall classification performance. The\nresults are evaluated using mean average of precision, recall and F1 score. The\nproposed approach achieved 91%, 95% and 92% on precision, recall and F1 score,\nrespectively, for truss detection using the ResNet101 with dataset augmentation\nutilising Salt-and-Pepper noise; and 83%, 53% and 65% on precision, recall and\nF1 score, respectively, for truss detection using the ResNet50 with dataset\naugmentation utilising Poisson noise.\n", "date": "2022-07-06", "categories": []}
{"id": "2207.11228", "title": "Classifying Crop Types using Gaussian Bayesian Models and Neural\n  Networks on GHISACONUS USGS data from NASA Hyperspectral Satellite Imagery", "abstract": "  Hyperspectral Imagining is a type of digital imaging in which each pixel\ncontains typically hundreds of wavelengths of light providing spectroscopic\ninformation about the materials present in the pixel. In this paper we provide\nclassification methods for determining crop type in the USGS GHISACONUS data,\nwhich contains around 7,000 pixel spectra from the five major U.S. agricultural\ncrops (winter wheat, rice, corn, soybeans, and cotton) collected by the NASA\nHyperion satellite, and includes the spectrum, geolocation, crop type, and\nstage of growth for each pixel. We apply standard LDA and QDA as well as\nBayesian custom versions that compute the joint probability of crop type and\nstage, and then the marginal probability for crop type, outperforming the\nnon-Bayesian methods. We also test a single layer neural network with dropout\non the data, which performs comparable to LDA and QDA but not as well as the\nBayesian methods.\n", "date": "2022-07-21", "categories": []}
{"id": "2207.11238", "title": "Improved lightweight identification of agricultural diseases based on\n  MobileNetV3", "abstract": "  At present, the identification of agricultural pests and diseases has the\nproblem that the model is not lightweight enough and difficult to apply. Based\non MobileNetV3, this paper introduces the Coordinate Attention block. The\nparameters of MobileNetV3-large are reduced by 22%, the model size is reduced\nby 19.7%, and the accuracy is improved by 0.92%. The parameters of\nMobileNetV3-small are reduced by 23.4%, the model size is reduced by 18.3%, and\nthe accuracy is increased by 0.40%. In addition, the improved MobileNetV3-small\nwas migrated to Jetson Nano for testing. The accuracy increased by 2.48% to\n98.31%, and the inference speed increased by 7.5%. It provides a reference for\ndeploying the agricultural pest identification model to embedded devices.\n", "date": "2022-07-19", "categories": []}
{"id": "2207.07334", "title": "Computer Vision for Volunteer Cotton Detection in a Corn Field with UAS\n  Remote Sensing Imagery and Spot Spray Applications", "abstract": "  To control boll weevil (Anthonomus grandis L.) pest re-infestation in cotton\nfields, the current practices of volunteer cotton (VC) (Gossypium hirsutum L.)\nplant detection in fields of rotation crops like corn (Zea mays L.) and sorghum\n(Sorghum bicolor L.) involve manual field scouting at the edges of fields. This\nleads to many VC plants growing in the middle of fields remain undetected that\ncontinue to grow side by side along with corn and sorghum. When they reach\npinhead squaring stage (5-6 leaves), they can serve as hosts for the boll\nweevil pests. Therefore, it is required to detect, locate and then precisely\nspot-spray them with chemicals. In this paper, we present the application of\nYOLOv5m on radiometrically and gamma-corrected low resolution (1.2 Megapixel)\nmultispectral imagery for detecting and locating VC plants growing in the\nmiddle of tasseling (VT) growth stage of cornfield. Our results show that VC\nplants can be detected with a mean average precision (mAP) of 79% and\nclassification accuracy of 78% on images of size 1207 x 923 pixels at an\naverage inference speed of nearly 47 frames per second (FPS) on NVIDIA Tesla\nP100 GPU-16GB and 0.4 FPS on NVIDIA Jetson TX2 GPU. We also demonstrate the\napplication of a customized unmanned aircraft systems (UAS) for spot-spray\napplications based on the developed computer vision (CV) algorithm and how it\ncan be used for near real-time detection and mitigation of VC plants growing in\ncorn fields for efficient management of the boll weevil pests.\n", "date": "2022-07-15", "categories": []}
{"id": "2207.06673", "title": "Detecting Volunteer Cotton Plants in a Corn Field with Deep Learning on\n  UAV Remote-Sensing Imagery", "abstract": "  The cotton boll weevil, Anthonomus grandis Boheman is a serious pest to the\nU.S. cotton industry that has cost more than 16 billion USD in damages since it\nentered the United States from Mexico in the late 1800s. This pest has been\nnearly eradicated; however, southern part of Texas still faces this issue and\nis always prone to the pest reinfestation each year due to its sub-tropical\nclimate where cotton plants can grow year-round. Volunteer cotton (VC) plants\ngrowing in the fields of inter-seasonal crops, like corn, can serve as hosts to\nthese pests once they reach pin-head square stage (5-6 leaf stage) and\ntherefore need to be detected, located, and destroyed or sprayed . In this\npaper, we present a study to detect VC plants in a corn field using YOLOv3 on\nthree band aerial images collected by unmanned aircraft system (UAS). The\ntwo-fold objectives of this paper were : (i) to determine whether YOLOv3 can be\nused for VC detection in a corn field using RGB (red, green, and blue) aerial\nimages collected by UAS and (ii) to investigate the behavior of YOLOv3 on\nimages at three different scales (320 x 320, S1; 416 x 416, S2; and 512 x 512,\nS3 pixels) based on average precision (AP), mean average precision (mAP) and\nF1-score at 95% confidence level. No significant differences existed for mAP\namong the three scales, while a significant difference was found for AP between\nS1 and S3 (p = 0.04) and S2 and S3 (p = 0.02). A significant difference was\nalso found for F1-score between S2 and S3 (p = 0.02). The lack of significant\ndifferences of mAP at all the three scales indicated that the trained YOLOv3\nmodel can be used on a computer vision-based remotely piloted aerial\napplication system (RPAAS) for VC detection and spray application in near\nreal-time.\n", "date": "2022-07-14", "categories": []}
{"id": "2207.03638", "title": "A Support Vector Model of Pruning Trees Evaluation Based on OTSU\n  Algorithm", "abstract": "  The tree pruning process is the key to promoting fruits' growth and improving\ntheir productions due to effects on the photosynthesis efficiency of fruits and\nnutrition transportation in branches. Currently, pruning is still highly\ndependent on human labor. The workers' experience will strongly affect the\nrobustness of the performance of the tree pruning. Thus, it is a challenge for\nworkers and farmers to evaluate the pruning performance. Intended for a better\nsolution to the problem, this paper presents a novel pruning classification\nstrategy model called \"OTSU-SVM\" to evaluate the pruning performance based on\nthe shadows of branches and leaves. This model considers not only the available\nilluminated area of the tree but also the uniformity of the illuminated area of\nthe tree. More importantly, our group implements OTSU algorithm into the model,\nwhich highly reinforces robustness of the evaluation of this model. In\naddition, the data from the pear trees in the Yuhang District, Hangzhou is also\nused in the experiment. In this experiment, we prove that the OTSU-SVM has good\naccuracy with 80% and high performance in the evaluation of the pruning for the\npear trees. It can provide more successful pruning if applied into the orchard.\nA successful pruning can broaden the illuminated area of individual fruit, and\nincrease nutrition transportation from the target branch, dramatically\nelevating the weights and production of the fruits.\n", "date": "2022-07-08", "categories": []}
{"id": "2206.15042", "title": "Automated Wheat Disease Detection using a ROS-based Autonomous Guided\n  UAV", "abstract": "  With the increase in world population, food resources have to be modified to\nbe more productive, resistive, and reliable. Wheat is one of the most important\nfood resources in the world, mainly because of the variety of wheat-based\nproducts. Wheat crops are threatened by three main types of diseases which\ncause large amounts of annual damage in crop yield. These diseases can be\neliminated by using pesticides at the right time. While the task of manually\nspraying pesticides is burdensome and expensive, agricultural robotics can aid\nfarmers by increasing the speed and decreasing the amount of chemicals. In this\nwork, a smart autonomous system has been implemented on an unmanned aerial\nvehicle to automate the task of monitoring wheat fields. First, an image-based\ndeep learning approach is used to detect and classify disease-infected wheat\nplants. To find the most optimal method, different approaches have been\nstudied. Because of the lack of a public wheat-disease dataset, a custom\ndataset has been created and labeled. Second, an efficient mapping and\nnavigation system is presented using a simulation in the robot operating system\nand Gazebo environments. A 2D simultaneous localization and mapping algorithm\nis used for mapping the workspace autonomously with the help of a\nfrontier-based exploration method.\n", "date": "2022-06-30", "categories": []}
{"id": "2206.13406", "title": "Explicitly incorporating spatial information to recurrent networks for\n  agriculture", "abstract": "  In agriculture, the majority of vision systems perform still image\nclassification. Yet, recent work has highlighted the potential of spatial and\ntemporal cues as a rich source of information to improve the classification\nperformance. In this paper, we propose novel approaches to explicitly capture\nboth spatial and temporal information to improve the classification of deep\nconvolutional neural networks. We leverage available RGB-D images and robot\nodometry to perform inter-frame feature map spatial registration. This\ninformation is then fused within recurrent deep learnt models, to improve their\naccuracy and robustness. We demonstrate that this can considerably improve the\nclassification performance with our best performing spatial-temporal model\n(ST-Atte) achieving absolute performance improvements for\nintersection-over-union (IoU[%]) of 4.7 for crop-weed segmentation and 2.6 for\nfruit (sweet pepper) segmentation. Furthermore, we show that these approaches\nare robust to variable framerates and odometry errors, which are frequently\nobserved in real-world applications.\n", "date": "2022-06-27", "categories": []}
{"id": "2206.11920", "title": "Agriculture-Vision Challenge 2022 -- The Runner-Up Solution for\n  Agricultural Pattern Recognition via Transformer-based Models", "abstract": "  The Agriculture-Vision Challenge in CVPR is one of the most famous and\ncompetitive challenges for global researchers to break the boundary between\ncomputer vision and agriculture sectors, aiming at agricultural pattern\nrecognition from aerial images. In this paper, we propose our solution to the\nthird Agriculture-Vision Challenge in CVPR 2022. We leverage a data\npre-processing scheme and several Transformer-based models as well as data\naugmentation techniques to achieve a mIoU of 0.582, accomplishing the 2nd place\nin this challenge.\n", "date": "2022-06-23", "categories": []}
{"id": "2206.01734", "title": "Using UAS Imagery and Computer Vision to Support Site-Specific Weed\n  Control in Corn", "abstract": "  Currently, weed control in a corn field is performed by a blanket application\nof herbicides that do not consider spatial distribution information of weeds\nand also uses an extensive amount of chemical herbicides. To reduce the amount\nof chemicals, we used drone-based high-resolution imagery and computer-vision\ntechniques to perform site-specific weed control in corn.\n", "date": "2022-06-02", "categories": []}
{"id": "2606.22439", "title": "Curvature-aware 3D length estimation of greenhouse cucumbers using RGB-D imaging and cubic spline arc-length integration", "abstract": "Commercial greenhouse cucumber production is graded by fruit length, which drives harvest scheduling, labour allocation, and logistics. Manual measurement with thread or caliper is accurate but infeasible at commercial scale. This paper presents CucumberVision, a non-contact length estimation framework using an Intel RealSense D435 RGB-D camera. A YOLO26n instance segmentation model locates cucumbers, and SAM (ViT-B backbone) refines each detection to a pixel-precise mask. Five methods are evaluated under matched conditions: (M1) a dominant-axis skeleton scan-line baseline; (M2) PCA on the bounding-box depth point cloud; (M3) SAM mask with medial-axis skeletonisation; (M4) a hybrid keypoint-guided approach using a YOLO26-pose model predicting five anatomical landmarks (KP0--KP4) with piecewise 3D arc-length; and (M5) a novel medial arc spline method fitting a cubic spline through the 3D medial axis of the SAM mask and computing arc length by trapezoidal integration -- the first such application to elongated vegetable measurement. All methods share five-frame burst depth averaging, colour-stream intrinsic alignment, and adaptive method selection with cascading fallbacks ensuring 100% coverage. A benchmark of 48 captures across seven cucumbers in three size categories (small ~8 cm, medium ~13 cm, large ~25 cm) with thread-based ground truth establishes a significant accuracy hierarchy: M1 (MAPE 9.68%) > M2 (5.31%) > M4 (5.51%) > M3 (5.82%) > M5 (4.13%). M5 significantly outperforms all competitors at Bonferroni-corrected alpha=0.0125. A secondary contribution is identifying a 12--18% length underestimation caused by using depth-stream rather than colour-stream intrinsics after rs.align(rs.stream.color) -- an under-reported error source. The complete system is released open source and runs in real time on a single consumer-grade GPU.", "date": "2026-06-21", "categories": []}
{"id": "2606.23441", "title": "Cross-Architectural Mixture-of-Experts with Adaptive Soft Routing for Plant Leaf Disease Classification", "abstract": "Plant leaf disease classification is crucial for crop protection and precision agriculture but remains challenging under complex backgrounds, illumination variations, and severe class imbalance. Moreover, single-architecture models often fail to effectively capture both local and global representations. To address these challenges, this study proposes an adaptive soft Mixture-of-Experts (MoE) framework with cross-architectural routing that integrates EfficientNet-B0, DenseNet-121, and Swin-Tiny to exploit complementary multi-scale, local, and global features. A soft gating mechanism dynamically assigns input-dependent expert weights, while a two-stage refinement training strategy improves optimization stability and generalization. Experiments on a highly imbalanced potato leaf disease dataset achieve 91.68% recall and 92.62% F1-score, surpassing the strongest individual expert by 5.91% and 5.03%, respectively. Additional evaluations on durian and sesame leaf disease datasets yield F1-scores of 94.03% and 97.04%, demonstrating robust cross-dataset generalization and the potential of the proposed framework for reliable real-world crop health monitoring", "date": "2026-06-22", "categories": []}
{"id": "2606.26165", "title": "Predicting Fruit Quality with a Hybrid Machine Learning and Image Processing Approach", "abstract": "Fruit spoilage is a significant issue in agriculture, leading to substantial economic losses. Addressing this, our study introduces a hybrid approach combining image processing and deep learning to assess fruit freshness. We developed an image processing algorithm that quantifies spoilage on a scale from 0 (fully fresh) to 100 (fully rotten). Alongside, we trained a convolutional neural network (CNN) to perform binary classification (fresh or rotten) using a large dataset of fruit images. The outcomes of both methods were synthesized using logistic regression to enhance the accuracy of freshness predictions. Subsequently, this logistic regression model was utilized to enable the image processing algorithm to provide binary classification based on its percentage output, thus eliminating the need for the CNN in real-time applications. Our approach, which does not require high computational resources, achieved real-time performance and was validated with over 90% accuracy on a dataset comprising apples and oranges. The primary limitation lies in the requirement for fruits to be isolated on a background that must be either white or transparent, suggesting future improvements could include advanced segmentation models to automate background removal. This study's results highlight the potential of integrating simple image processing techniques with machine learning to provide practical solutions in the agricultural sector.", "date": "2026-06-24", "categories": []}
{"id": "2606.26828", "title": "Learning Adversarial Augmentation Policies for Robust Garlic Seedling Detection", "abstract": "Accurate seedling detection during early growth stages is essential for timely replanting and effective crop management in precision agriculture. However, existing studies are mostly evaluated under relatively stable imaging conditions, such as UAV imagery or greenhouse environments, leaving robust detection under severe and spatially heterogeneous illumination in ground-based outdoor monitoring insufficiently explored. In addition, many illumination-robust detection methods rely on additional enhancement or feature-extraction modules, which increase inference-time overhead and are not tailored to seedling detection and downstream missing seedling localization. To address these gaps, we construct a new garlic seedling dataset captured using a ground-based monitoring platform under real outdoor field conditions with highly variable illumination. We further propose an illumination-robust seedling detection framework based on adversarial augmentation policy learning. The proposed method jointly optimizes a stochastic augmentation policy agent and an object detector, enabling the detector to learn robust representations under challenging visual conditions. A structural penalty is introduced to prevent unrealistic distortions while encouraging challenging augmentations during training. Extensive experiments show that the proposed approach achieves an AP$_{50}$ of 91.6%, improving the baseline by 0.9 percentage points and outperforming the previous best-performing method by 0.2 percentage points. For downstream missing seedling localization, it achieves 75.0% precision and a 67.0% F1-score, improving the baseline by 4.8 and 2.0 percentage points, respectively. These results demonstrate the effectiveness of the proposed framework for practical ground-based agricultural monitoring under complex outdoor lighting conditions without additional inference-time computational overhead.", "date": "2026-06-25", "categories": []}
{"id": "2606.29664", "title": "Benchmarking Geospatial Foundation Models for Agriculture Applications", "abstract": "Geospatial foundation models pretrained on satellite imagery promise broad generalization across remote sensing tasks and regions, but their geographic transferability has not been systematically tested, especially in agriculture applications. This paper presents a controlled benchmark that evaluates three models, Prithvi, SpectralGPT, and SatMAE, on multi-temporal crop segmentation and change detection across four U.S. states, Iowa, North Carolina, California, and Minnesota. By assigning each train, validation, and test split to a separate region, we measure how well each model transfers to land it has not seen. All three degrade sharply under regional distribution shift, predicting only the most common crops while missing rare ones. We further find that fitting these models to a shared input format affects each one differently, which complicates direct architectural comparison. These results expose key limitations of current geospatial foundation models for agriculture and point to region aware evaluation as a necessary standard.", "date": "2026-06-29", "categories": []}