Datasets:

Project-AgML
/

banana_bunch_detection

dataset_info:
  features:
    - name: image
      dtype: image
    - name: objects
      struct:
        - name: bbox
          list:
            list: float64
        - name: categories
          list:
            class_label:
              names:
                '0': '0'
  splits:
    - name: train
      num_bytes: 19797911338
      num_examples: 2179
  download_size: 17716781372
  dataset_size: 19797911338
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: cc-by-4.0
task_categories:
  - object-detection
size_categories:
  - 1K<n<10K

Banana Bunch Detection

A dataset for object detection of Banana bunches. The dataset contains 2,179 images with 3,718 bounding box annotations across 1 category.

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

Citation

@article{baglat2025multi,
  title={A multi-stage dataset for banana bunch detection and harvesting decision support},
  author={Baglat, Preety and Mendon{\c{c}}a, F{\'a}bio and Mostafa, Shanawaz and Morgado-Dias, Fernando},
  journal={Data in Brief},
  pages={112337},
  year={2025},
  publisher={Elsevier}
}

Baglat, P., Hayat, A., Mostafa, S. S., Mendonça, F.& Morgado Dias, F. (2025). Banana Bunch Dataset: Multi-Field Acquisition with Various Environmental Conditions [Graphic]. Zenodo. https://doi.org/10.5281/zenodo.15642838