RSANet: Relative-sequence quality assessment network for gait recognition in the wild

Guozhen Peng, Yunhong Wang, Shaoxiong Zhang, Rui Li, Yuwei Zhao, Annan Li

Abstract: Highlights•We design a RSANet to estimate the silhouette quality in a relative-sequence manner.•The proposed AFSM calculates AFS sequence and 3D AFS sequence to weight silhouette.•The proposed MTB extends temporal receptive field without parameter increases.•The proposed model achieves superior performance on in-the-wild gait datasets.