Review on CNN-based violent detection method
Guanjie Liang · Applied and Computational Engineering · 2023
Violent behaviors seriously endanger people’s life and property safety, and also undermine social stability and development. In order to monitor the occurrence of violent behavior, the monitoring system plays a vitally important role. As the surveillance is used widely in daily life and video data is growing rapidly, it is increasingly unrealistic to manually detect violent behavior in surveillance, because it will consume excessive manpower. Therefore, it is necessary to establish an automated system for detecting violent behavior. And the convolutional neural network (CNN) plays a leading role in automatic detection. This paper has done a lot of research to study and understand the development of CNN-based violent behavior recognition. First, the original CNN is introduced and the related work is mentioned; Then, the two different improvement paths on CNN, 2DCNN+RNN and 3DCNN, are utilized to enhance the accuracy of violence detection or reduce the calculation. And finally, the discussion about advantages and disadvantages of these two improvement paths is shown in this paper and conclusion is presented. At present, the development of potential of CNN remains to be exploited and the development of CNN is expected.