Polyp-Size: A Precisely Annotated Endoscopic Dataset for AI-Assisted Polyp Sizing

The Polyp-Size dataset consists of 42 high-resolution white-light colonoscopy videos with vernier caliper-validated polyp measurements, accompanied by comprehensive metadata including patient demographics, equipment specifications, precise polyp sizes, Paris classifications, anatomical locations, and histological findings. The videos are standardized to 1000×1170 pixels at 30fps, with each video averaging 9.76 seconds in duration, making this dataset uniquely valuable for developing AI-based polyp size estimation systems.