Hands, face and joints for multi-modal human-action temporal segmentation and recognition | IEEE Conference Publication | IEEE Xplore