Datasets:

Project-AgML
/

salma_leaf_segmentation

dataset_info:
  features:
    - name: image
      dtype: image
    - name: mask
      dtype: image
    - name: species
      dtype: string
    - name: source_set
      dtype: string
  splits:
    - name: train
      num_bytes: 354176321
      num_examples: 3674
  download_size: 354200623
  dataset_size: 354176321
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: cc-by-4.0
task_categories:
  - image-segmentation
size_categories:
  - 1K<n<10K

Salma Leaf Segmentation

The SALMA benchmark: single-leaf crops of scanned (600 dpi) and photographed leaves with human-delineated leaf/background masks, used to validate a machine-learning tool for leaf area and morphology. The main set covers 3,332 leaves of 64 New Zealand species spanning most of the global range of leaf sizes; two smaller sets hold 130 low-contrast photographs of four alpine species and 212 morphologically complex or damaged leaves of six species. Species and source set are kept as columns. The dataset contains 3,674 images with pixel-level mask annotations.

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

Citation

@article{shabanov2026salma,
  title={SALMA: A machine learning tool for precise leaf morphology measurements},
  author={Shabanov, Ilya and Deslippe, Julie R and Lensen, Andrew},
  journal={Ecological Informatics},
  volume={93},
  pages={103592},
  year={2026},
  publisher={Elsevier}
}

The dataset itself can be cited as:

Shabanov, I., Deslippe, J., & Lensen, A. (2025). Datasets for SALMA: A Machine Learning Tool for Precise Leaf Morphology Measurements of Small Leaves (Version 1.0) [Dataset]. Zenodo. https://doi.org/10.5281/ZENODO.18090525

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