Datasets:

Project-AgML
/

cocoa_tree_point_cloud_segmented

dataset_info:
  features:
    - name: plant_id
      dtype: string
    - name: points
      dtype:
        array2_d:
          shape:
            - -1
            - 3
          dtype: float32
    - name: colors
      dtype:
        array2_d:
          shape:
            - -1
            - 3
          dtype: float32
    - name: label
      dtype: int64
  splits:
    - name: train
      num_bytes: 708179899
      num_examples: 85
  download_size: 564612484
  dataset_size: 708179899
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: cc-by-nc-sa-4.0
task_categories:
  - image-classification
size_categories:
  - n<1K

Cocoa Tree Point Cloud Segmented

This dataset provides real LiDAR point cloud data of cocoa trees in a field environment in Cameroon, collected for crop segmentation applications within agroforestry systems. Captured using a ground-based Leica ScanStation C10 during August 2019, it delivers high-resolution structural information of cocoa tree canopies for agricultural monitoring research. The dataset contains 85 images across 3 classes: full, leaf, wood.
Images per class:

full: 48
leaf: 9
wood: 28

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

Citation

@article{peynaud2024terrestrial,
  title={Terrestrial LiDAR point cloud dataset of cocoa trees grown in agroforestry systems in Cameroon},
  author={Peynaud, Emilie and Momo Takoudjou, Stéphane},
  journal={Data in Brief},
  volume={53},
  pages={110108},
  year={2024},
  publisher={Elsevier}
}

Momo Takoudjou, S., & Peynaud, E. (2021). Cocoa tree point clouds obtained by terrestrial Lidar scanning in agroforestry systems in Cameroon (Version V3) [dataset]. CIRAD Dataverse. https://doi.org/doi:10.18167/DVN1/5HZB1F

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