Action recognition by learning mid-level motion features.

Alireza Fathi, Greg Mori

Browse the full CVPR paper archive.