Datasets:

Project-AgML
/

AppleTreeDataset_intertwined_segmentation

dataset_info:
  features:
    - name: image
      dtype: image
    - name: mask
      dtype: image
    - name: split
      dtype: string
    - name: capture_date
      dtype: string
    - name: camera
      dtype: string
    - name: tree_id
      dtype: string
  splits:
    - name: train
      num_bytes: 3042168098
      num_examples: 150
  download_size: 3042238144
  dataset_size: 3042168098
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: cc-by-4.0
task_categories:
  - image-segmentation
size_categories:
  - n<1K

Appletreedataset Intertwined Segmentation

This dataset provides real RGB images of intertwined apple trees in a field environment, capturing complex branch and fruit structures during the 2021 harvest season. Collected using a handheld iPhone 6 at a South Korean agricultural research facility, it offers a naturalistic representation of outdoor crop conditions for semantic segmentation tasks. The dataset contains 150 images with pixel-level mask annotations.

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{la2023deep,
  title={Deep Learning-Based Segmentation of Intertwined Fruit Trees for Agricultural Tasks},
  author={La, Young-Jae and Seo, Dasom and Kang, Junhyeok and Kim, Minwoo and Yoo, Tae-Woong and Oh, Il-Seok},
  journal={Agriculture},
  volume={13},
  pages={2097},
  year={2023},
  publisher={MDPI}
}

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