PROCEEDINGS OF THE ITE ANNUAL CONVENTION
Online ISSN : 2424-2292
Print ISSN : 1343-1846
ISSN-L : 1343-1846
2013
Session ID : 1-4
Conference information

1-4 Human action recognition using a spatio-temporal pairwise feature representation
Masaki TAKAHASHIMasahide NAEMURAMahito FUJIIJames J. LITTLE
Author information
CONFERENCE PROCEEDINGS FREE ACCESS

Details
Abstract
A new feature-representation method for action recognition in video sequences, which conveys spatio-temporal information of local motion features, is proposed. The method considers the co-occurrence of camera motions and human actions. It was experimentally confirmed that the proposed method can robustly detect specific human actions in sports broadcast videos.
Content from these authors
© 2013 The Institute of Image Information and Television Engineers
Previous article Next article
feedback
Top