Multi-stream feature refinement network for human object interaction detection

Zhanpeng Shao, Zhongyan Hu, Jianyu Yang, Youfu Li

Abstract: Highlights•A simple yet effective multi-stream network for HOI detection is proposed.•Visual features are organized to explicitly receive key cues for feature refinement.•We prove that how the features are refined matters more than the adopted features.•Experiments show that the network is simpler but performs better than those SOTAs.