P-CNN: Pose-Based CNN Features for Action Recognition.

Guilhem Chron, Ivan Laptev, Cordelia Schmid

Browse the full ICCV paper archive.