Abnormal Behavior Recognition System based on Improved CRNN Model

Abnormal behavior recognition technology based on surveillance video has been an important direction of current research, with high research value and application demand. Due to the complexity and variability of crowd movement and external environment, the recognition of abnormal behavior is quite challenging, and there is still a need for further research on the recognition technology of human abnormal behavior in surveillance video. In this paper, we proposed an adaptive video frame extraction method, and after a series of experiments and comparisons with former methods, our model achieved 86.78% accuracy on the UCF-Crime dataset and 95.2% accuracy in the subsequent system development of the algorithm model for segment detection and localization of surveillance videos. And the maximum speed of recognition can reach 543. 7FPS. Data shows that the algorithm and system we designed can achieve better performance on Abnormal behavior recognition.