| Paper: | WP-L5.2 |
| Session: | Motion Detection and Estimation III |
| Time: | Wednesday, September 19, 14:50 - 15:10 |
| Presentation: |
Lecture
|
| Title: |
DOMINANT SETS-BASED ACTION RECOGNITION USING IMAGE SEQUENCE MATCHING |
| Authors: |
Qingdi Wei; National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences | | |
| | Weiming Hu; National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences | | |
| | Xiaoqin Zhang; National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences | | |
| | Guan Luo; National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences | | |
| Abstract: |
Action recognition is one of the most active research fields in computer vision. In this paper, we propose a novel method for classifying human actions from a series of image sequences containing certain actions.. Human action in image sequences can be recognized from a time-varying contour of human body. We first extract shape context of each contour to form the feature space. Then the dominant sets approach is used for feature clustering and classification to get sequences. Finally, we use a smoothing algorithm upon the label sequences to recognize human actions . The proposed dominant sets-based approach has been tested in comparison to three classical methods: K-means, mean shift, and Fuzzy-Cmean. Experimental results show that the dominant sets-based approach achieves the best recognition performance. Moreover, our method is robust to non-rigid deformations, significant scale changes, high action irregularities, and low quality video. |
