Datasets:

Project-AgML
/

bruised_vegetable_classification

dataset_info:
  features:
    - name: image
      dtype: image
    - name: label
      dtype:
        class_label:
          names:
            '0': Damaged
            '1': Fresh
            '2': Severely Damaged
    - name: crop_type
      dtype: string
    - name: plant_id
      dtype: int64
  splits:
    - name: train
      num_bytes: 10632989700
      num_examples: 4464
  download_size: 10744868382
  dataset_size: 10632989700
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: cc-by-4.0
task_categories:
  - image-classification
size_categories:
  - 1K<n<10K

Bruised Vegetable Classification

A dataset for classification of Bruised Vegetable Classification. The dataset contains 4,464 images across 3 classes.

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

Citation

@article{samanta2025nature,
  title={Nature's best vs. bruised: A veggie edibility evaluation database},
  author={Samanta, Bidisha and Banerjee, Sriparna and Das, Ranadhir and Chaudhuri, Sheli Sinha and Djemal, Khalifa and Feiz, Amir Ali},
  journal={Data in Brief},
  volume={60},
  pages={111483},
  year={2025},
  publisher={Elsevier}
}

Samanta, Bidisha ; Banerjee, Sriparna; Das, Ranadhir; Sinha Chaudhuri, Sheli; Djemal, Khalifa (2024), “Nature's Best vs. Bruised: A Veggie Evaluation”, Mendeley Data, V2, doi: 10.17632/b2mvj3kjfx.2

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