Title:EgoSampling: Fast-Forward and Stereo for Egocentric Videos

Abstract:Use of egocentric cameras is on the rise, with devices like GoPro cameras and Google Glass gaining popularity. The videos captured by such cameras are long, boring, and difficult to watch from start to end. Fast forwarding by frame sampling is a natural choice for faster video browsing. However, in egocentric videos the shake introduced by natural head motion gets accentuated, making the fast forwarded video useless. We propose EgoSampling, a novel approach for frame sampling which enables a more stable fast forward for egocentric videos. Frame sampling is formulated as an energy minimization problem, whose optimal solution can be found in polynomial time.
Egocentric video taken while walking suffers from the left-right movement of the head as the body weight shifts from one leg to another. We show that this drawback can be turned into a feature: Stereo video can be created by sampling the frames from the left most and right most head positions of each step, forming approximate stereo-pairs.