3D Human Action Representation Learning via Cross-View Consistency Pursuit | IEEE Conference Publication | IEEE Xplore