A temporal densely connected recurrent network for event-based human pose estimation

Zhanpeng Shao, Xueping Wang, Wen Zhou, Wuzhen Wang, Jianyu Yang, Youfu Li

Abstract: Highlights•A temporal densely connected recurrent network for event-based HPE is proposed.•The network is able to alleviate the problem of incomplete signals in event frames.•A multimodal event-based pose dataset is collected as a new challenging benchmark.•This research can help unlock the potentials of event cameras being applied in HPE.