Space-Time Neighborhood Based Hierarchical Descriptor for Action Recognition
Recent work shows interest-point-based representation is greatly popular in action recognition, due to their simple implementation and good reliability. The neighborhood information of local descriptors usually improves the recognition accuracy. Taking inspiration from this observation, we propose a novel hierarchical neighborhood descriptor for action recognition. At low level, we propose the compound appearance and motion descriptor which describes the feature of neighboring interest points, rather than a single space-time interest point. At high level, another new neighborhood based descriptor is proposed to describe the spatial distribution of neighboring interest points. For classification, we apply multi-channel nonlinear SVM based on the hierarchical vocabulary. Experiments validate that our method achieves the state-of-the-art results on two benchmark datasets.
Interest point space-time neighborhood hierarchical structure.
Haoran Wang Chunfeng Yuan Weiming Hu Changyin Sun
School of Automation Southeast UniversityNanjing, China National Laboratory of Pattern RecognitionInstitute of Automation, CASBeijing, China School of Automation Southeast University Nanjing, China
国际会议
北京
英文
95-99
2011-11-28（万方平台首次上网日期，不代表论文的发表时间）