TSdetector: Temporal-Spatial self-correction collaborative learning for colonoscopy video detection

Open Webpage

Kai-Ni Wang, Haolin Wang, Guang-Quan Zhou, Yangang Wang, Ling Yang, Yang Chen, Shuo Li

Published: 2025, Last Modified: 27 Feb 2026Medical Image Anal. 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•An innovative Temporal–Spatial self-correcting Collaborative Learning network for polyp video detection.•An efficient global time-aware convolution and hierarchical queue integration mechanism is designed.•Position-aware clustering is proposed, a new method that exploits the relationship between candidate boxes to adjust confidence adaptively.•State-of-the-art results were achieved on the largest public polyp video dataset, with a polyp detection rate of 95.30%.