Movement Human Actions Recognition Based on Machine Learning

Honghua Xu, Li Li, Ming Fang, Fengrong Zhang


In this paper, the main technologies of foreground detection, feature description and extraction, movement behavior classification and recognition were introduced. Based on optical flow for movement objects detection, optical flow energy image was put forward for movement feature expression and region convolutional neural networks was adopt to choose features and decrease dimension. Then support vector machine classifier was trained and used to classify and recognize actions. After training and testing on public human actions database, the experiment result showed that the method could effectively distinguish human actions and significantly improved the recognition accuracy of human actions. And for the different situations of camera lens drawing near, pulling away or slight movement of camera, the solution had recognition effect as well. At last, this scheme was applied to intelligent video surveillance system, which was used to identify abnormal behavior and alarm. The abnormal behaviors of faint, smashing car, robbery and fighting were defined in the system. In running of the system, it obtained satisfactory recognition results.


optical flow; support vector machines; convolutional neural networks

Full Text:


International Journal of Online and Biomedical Engineering (iJOE) – eISSN: 2626-8493
Creative Commons License
Scopus logo Clarivate Analyatics ESCI logo IET Inspec logo DOAJ logo DBLP logo EBSCO logo Ulrich's logo Google Scholar logo MAS logo