Hemorrhage Recognition in Minimally Invasive Surgery: A Multi-Source Pixel-Level Annotated Dataset

The Laparoscopic Cholecystectomy Hemorrhage Segmentation Dataset (LC-HemSeg-30) contains 21,396 expertly validated frames, capturing both common and rare bleeding events in diverse surgical settings. The dataset uniquely highlights the dynamic evolution of hemorrhagic patterns throughout critical operative phases, providing a high-fidelity foundation for algorithm development, workflow analysis, and real-time surgical risk prediction. (This dataset is also referred to as the Southern Medical University - Zhujiang Hospital - Cholecystectomy Hemorrhage - 01 Dataset, SMU-ZJ-LC-Hem-01).