Abstract
Human action sequences can be considered as nonlinear dynamic manifolds in image frames space. In this paper, a novel manifold embedding method, Maximum Temporal Inter-class Dissimilarity (MTID), is proposed for human action recognition, which is based on the framework of Locality Preserving Projections (LPP). Being different from LPP whose goal is to minimize the intra-class distance in local neighborhood, MTID can make best of both the class label information and the temporal information to maximize the inter-class distance in local neighborhood, Namely, focusing on maximizing the dissimilarity between frames that are similar in appearance but are from different classes. At last the Nearest Neighbors classifier based on Hausdorff distance is introduced for recognition. The experimental results demonstrate the effectiveness of the proposed method for human action recognition.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
