CONVOLUTIONAL AUTOENCODER FOR ANOMALY DETECTION IN CROWDED SCENES
Mohamed Ashraf Ali, Maryam Nabil Al-Berry, Zaki Taha · Journal of Southwest Jiaotong University · 2021
Monitoring abnormal events in a crowded scene is essential these days, especially with the increase in surveillance cameras in most, if not all, places. This has made the field of computer vision an active field of research in recent periods. Identifying abnormal events in a crowded scene as soon as possible is essential, especially from the security side. It is a cumbersome and challenging task for humans because there are many surveillance cameras and overcrowding. Computer vision can solve this problem and get accurate and high results. This paper proposes a convolutional neural network architecture for anomaly detection in videos. The proposed convolutional neural network model has been trained to recognize anomaly frames. The performance of the proposed network has been evaluated using Avenue and UCSD standard datasets designed to identify anomalies events in crowded scenes. Experimental results show that the proposed model outperforms state-of-the-art methods, achieving an AUC of 71.96% and 89.52% on Avenue and UCSD Pde2 datasets. A detailed analysis of the experiments used to choose the parameters is presented and compared with other methods.