Datasets:

Project-AgML
/

pea_seed_viability_classification

dataset_info:
  features:
    - name: image
      dtype: image
    - name: label
      dtype:
        class_label:
          names:
            '0': dead
            '1': live
    - name: split
      dtype: string
  splits:
    - name: train
      num_bytes: 270035028
      num_examples: 3000
  download_size: 270022669
  dataset_size: 270035028
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

Pea Seed Viability Classification

This dataset provides real RGB images of pea seeds collected in a laboratory environment in Beijing, China. The images were captured using a fixed-position MV-EM120C CCD camera under controlled conditions, offering a standardized resource for agricultural computer vision research on seed viability assessment. The dataset contains 3,000 images across 2 classes: dead, live.
Images per class:

dead: 1,500
live: 1,500

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{men2026rapid,
  title={Rapid Seed Viability Detection Using Laser Speckle Weighted Generalized Difference with Improved Residual Networks},
  author={Men, Sen and Zhang, Junhao and Liu, Xinhong and Sun, Tianyi and Liu, Wei},
  journal={Agronomy},
  volume={16},
  pages={81},
  year={2026},
  publisher={MDPI}
}

The dataset itself can be cited as:

Sen Men. (2025). Rapid Seed Viability Detection Using Laser Speckle Weighted Generalized Difference with Improved Residual Networks [Dataset]. Zenodo. https://doi.org/10.5281/ZENODO.18041269

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