Combining hierarchical sparse representation with adaptive prompt for few-shot segmentation

Xiaoliu Luo, Ting Xie, Weisen Qin, Zhao Duan, Jin Tan, Taiping Zhang

Abstract: Highlights•The essential relation of targets in support and query images exhibits sparsity.•Propose the hierarchical sparse representation to extract essential relation.•Capture various category representations by an adaptive prompt meta-learner.•Extensive experiments on three datasets validate the effectiveness of HSRap.