Video Abnormal Event Detection Based on CNN and LSTM
Guangli Wu, Zhenzhou Guo, Leiting Li, Cheng‐Xiang Wang · 2020
Nowadays, people pay more and more attention to security issues. If we can warn the occurrence of abnormal events, we can greatly reduce the occurrence of public security issues. Therefore, the detection of abnormal events in videos is an important research topic. In this paper, an improved deep learning model TCNN-LSTM is obtained by combining CNN with LSTM. We carried out experiments on the standard dataset UMN and UCSD, and achieved good results, with the accuracy reaching 99.54% and 92.36% respectively, and F1 reaching 98.41% and 87.65% respectively, it can be seen that this model has strong generalization ability. At the same time, other models were compared on the UCSD dataset. The AUC of this model reached 97.00%, far higher than that of other models. Experimental results showed that the performance of our model was better than that of other models.