MoDeep: A Deep Learning Framework Using Motion Features for Human Pose Estimation.

Arjun Jain, Jonathan Tompson, Yann LeCun, Christoph Bregler

Browse the full ACCV paper archive.