Datasets:

Project-AgML
/

MH_Weed16_weed_detection

dataset_info:
  features:
    - name: image
      dtype: image
    - name: objects
      struct:
        - name: bbox
          list:
            list: float64
        - name: categories
          list:
            class_label:
              names:
                '0': '0'
                '1': '1'
                '2': '2'
                '3': '3'
                '4': '4'
                '5': '5'
                '6': '6'
                '7': '7'
                '8': '8'
                '9': '9'
                '10': '10'
                '11': '11'
                '12': '12'
                '13': '13'
                '14': '14'
  splits:
    - name: train
      num_bytes: 2759067041
      num_examples: 6656
  download_size: 2969930573
  dataset_size: 2759067041
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

MH Weed16 Weed Detection

A dataset for weed detection in Soybean fields. The dataset contains 6,656 images with 62,052 bounding box annotations across 15 categories.

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

Citation

@article{shinde2025indian,
  title={An Indian annotated weed dataset for computer vision tasks in precision farming},
  author={Shinde, Sayali and Attar, Vahida},
  journal={Data in Brief},
  volume={61},
  pages={111691},
  year={2025},
  publisher={Elsevier}
}

Shinde, Sayali; Attar, Dr. Vahida; Technological University Pune, COEP; Technology Innovation Hub, Indian Statistical Institute Kolkata, IDEAS (2025), “MH-Weed16:An Indian Multiclass Annotated Weed Dataset for Computer Vision Tasks ”, Mendeley Data, V2, doi: 10.17632/d3n3mgjjbv.2