Action Recognition by Dense Trajectories 密集轨迹的动作识别.pdfVIP

Action Recognition by Dense Trajectories † ¨ ‡ ‡ † Heng Wang Alexander Klaser Cordelia Schmid Cheng-Lin Liu †National Laboratory of Pattern Recognition ‡ LEAR, INRIA Grenoble, LJK Institute of Automation, Chinese Academy of Sciences Grenoble, France {hwang, liucl}@ {Alexander.Klaser, Cordelia.Schmid}@inria.fr Abstract Feature trajectories have shown to be efficient for rep- resenting videos. Typically, they are extracted using the KLT tracker or matching SIFT descriptors between frames. However, the quality as well as quantity of these trajecto- ries is often not sufficient. Inspired by the recent success of dense sampling in image classification, we propose an approach to describe videos by dense trajectories. We sam- ple dense points from each frame and track them based on displacement information from a dense optical flow field. Given a state-of-the-art optical flow algorithm, our trajec- tories are robust to fast irregular motions as well as shot boundaries. Additionally, dense trajectories cover the mo- tion information in videos well. We, also, investigate how to design descriptors to encode the trajectory information. We introduce a novel descriptor based on motion boundary histograms, which is robust to camera motion. This descriptor consistently outperforms other state-of-the-art descriptors, in particular in uncon- trolled realistic videos. We evaluate our video description KLT