Datasets:

Project-AgML
/

chili_leaf_curl_detection

dataset_info:
  features:
    - name: image
      dtype: image
    - name: objects
      struct:
        - name: bbox
          list:
            list: float64
        - name: categories
          list:
            class_label:
              names:
                '0': Healthy
                '1': Mites
                '2': Virus
    - name: split
      dtype: string
  splits:
    - name: train
      num_bytes: 89322357
      num_examples: 1419
  download_size: 92141473
  dataset_size: 89322357
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: apache-2.0
task_categories:
  - object-detection
size_categories:
  - 1K<n<10K

Chili Leaf Curl Detection

The dataset comprises real RGB images of chili leaves captured in a mixed agricultural environment at IARI, New Delhi, India, using a handheld smartphone. It depicts natural field conditions for object detection tasks focused on identifying disease symptoms related to chili leaf curl complex. The dataset contains 1,419 images with 12,205 bounding box annotations across 3 categories.

This dataset is indexed on https://project-agml.github.io/ as part of the AgML python library.

The original train/test/val split has been preserved in the split column.

Citation

@article{paul2025sca,
  title={SCA-MobiPlant: smartphone-deployed multistage attention fusion model for accurate field detection of chili leaf curl complex},
  author={Paul, Samrat and Emmadi, Venu and Sarkar, Mehulee and Das, Shubhajyoti and Roy, Anirban and Sinha, Parimal},
  journal={Plant Methods},
  volume={21},
  pages={138},
  year={2025},
  publisher={BioMed Central}
}

https://github.com/SamratPaul16/YOLO_DyE

This dataset was reformatted from its original format to match HuggingFace standards.