Datasets:

Project-AgML
/

soybean_variety_classification

configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: cc-by-4.0
task_categories:
  - image-classification
size_categories:
  - 10K<n<100K
dataset_info:
  features:
    - name: image
      dtype: image
    - name: label
      dtype:
        class_label:
          names:
            '0': Anjasmoro
            '1': Dega
            '2': Grobogan
  splits:
    - name: train
      num_bytes: 802073373
      num_examples: 12683
  download_size: 857306740
  dataset_size: 802073373

Soybean Variety Classification

A dataset for variety classification of Soybean seeds. The dataset contains 12,683 images across 3 classes: Anjasmoro, Dega, Grobogan.
Images per class:

Anjasmoro: 3,493
Dega: 4,490
Grobogan: 4,700

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

Citation

@article{hanafiah2026open,
  title={An Open Image Dataset of Indonesian Soybean Seed Varieties (Anjasmoro, Grobogan, DEGA-1) for Agricultural Research and Machine Learning Applications},
  author={Hanafiah, Diana Sofia and Alfi, Rahmatika and Lestami, Anggria and Purnamasari, Fanindia and Nurhasanah, Rossy and Syahraza, Muhammad Ariyo and Saputra, Muhammad Azis and Pane, Usman Ismail and Manurung, Steven and Buntoro, Yunus Tio and others},
  journal={Data in Brief},
  pages={112524},
  year={2026},
  publisher={Elsevier}
}

Syahraza, Muhammad Ariyo; Hanafiah, Diana Sofia ; Purnamasari, Fanindia; Nurhasanah, Rossy (2025), “Image Dataset of Local Indonesian Soybean Seed Varieties (Anjasmoro, Grobogan, and DEGA-1)”, Mendeley Data, V3, doi: 10.17632/c733bjz4m3.3