“Alternate Self-Dual Teaching: A New Framework for Weakly Supervised Semantic Segmentation Achieves State-of-the-Art Results”

https://pubmed.ncbi.nlm.nih.gov/38117621

The abstract describes a new end-to-end learning framework, alternate self-dual teaching (ASDT), that uses a dual-teacher single-student network and pulse width wave-like selection signal to generate high-quality pseudo segmentation masks for weakly supervised semantic segmentation, achieving new state-of-the-art results on PASCAL VOC 2012 and COCO-Stuff 10K datasets.