Abstract
Crowd behavior recognition under complex surveillance scenarios is a fundamental and important problem in crowd management application. In this paper, a comprehensive and specific overall-level dynamic attribute package is proposed by considering local pattern-related motion and group-level motion together to represent crowd movement. Curl and divergence map of normalized average motion vector field act as local pattern-related motion, which represents physical movement tendency of each particle. Group-level motion explores crowd interaction of inter-/intra-group, which focus on depicting crowd’s social dynamic property. The complementary characteristic of two motion representation in different level is analyzed and verified. Single frames in video clips and the corresponding dynamic attribute packages are sent into two-branch structured ConvNet, which can extract more discriminative spatial-temporal feature for behavior recognition. Experiment results conducted on CUHK dataset show that the proposed crowd behavior recognition framework outperforms than existing approaches and obtains the state-of-art performance.
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.
