Pea Seed Instance Segmentation — Demonstration Subset

This dataset is a five-image demonstration subset of the pea seed (Pisum sativum) instance segmentation dataset used in a manuscript currently under peer review: Unal, Y. and Gulzar, Y. (2026). Comparative Analysis of Automatic Mask Generation Methods for Agricultural Seed Instance Segmentation. Manuscript under review. The subset supports reproducibility of the evaluation pipeline described in the manuscript. It contains five top-view JPG photographs and the corresponding LabelMe polygon annotations, selected deterministically from the full 100-image dataset to cover the density spectrum from 40 to 542 instances per image. Contents:- images/ 5 top-view JPG photographs (4032 x 3024 pixels)- annotations/ 5 LabelMe polygon annotation files- metadata.csv Density tier, filename stem, ground-truth instance count- README.md Documentation The full 100-image dataset is available from the authors on reasonable request under a data-use agreement. Code repository: https://github.com/yavuz-unal/pea-seed-segmentation-benchmark