Overcoming Object Tracking Challenges for Abnormal Behavior Recognition

This paper is about developing an Intelligent Video System which takes advantage of the latest technology and research enhancement. The goal is to let such a system automatically recognize an abnormal human behavior. We realized it and implemented it with a several scenarios using the SVM model. The results were excellent as we reached the rate of less than 1% of false positives in certain conditions. Many datasets have been developed for such activities like the UT-Interaction dataset and the UCR Video web dataset. As far as our work was concerned, we used a dataset we developed within the university premises containing both simple and complex activities. The experimental results on real-time video streams show the feasibility of our system and its effectiveness in human activity tracking and recognition.