Ruukki VOS Dataset: Multi-Camera Video Object Segmentation Benchmark of Bulls

Ruukki VOS Dataset (2026-01-02) A multi-camera Video Object Segmentation (VOS) benchmark dataset for tracking bulls in a research barn at the Natural Resources Institute Finland (Luke). Challenge: Evaluates multi-camera tracking, re-identification across opposite camera views, and cross-view mask splitting for 5 bulls per pen. Annotations: Frame-by-frame instance segmentation masks with consistent IDs created using a human-in-the-loop tool (SAM3 + XMem). Data Split: Includes a 1.5-hour training set and a 1-hour validation set split into 4 distinct clips. Structure: Organized into train/ and validation/ directories. Each contains back_cam/ and front_cam/ subfolders with JPEGImages/ (JPG frames) and Annotations/ (PNG mask files).