Survey on Abnormal Event Detection and Signalling in Multiple Video Surveillance Scenes Using CNN

With the growth of urbanisation, the flow of people is increasing steadily every year. The likelihood of stampedes in public areas rises as a result of this trend. Monitoring the audience for the occurrence of odd circumstances and acting quickly to prevent them is necessary. Crowd analysis is typically done for purposes of security and public safety. It is difficult to continually handle for human operators to continuously scan the visual screens for any occurrence of interest. There has been a lot of study done in the area of anomalies identification in crowds by the computer vision and signal processing groups, which pushes researchers to design an autonomous system for doing so and assisting the operators. Recent attempts have been made to avoid using any labour-intensive hand-crafted feature extraction and processing methods by utilising deep learning models. There are shortcomings such ground truth availability, anomaly type, etc. that are highlighted in despite the extensive research and achievement in this area. The development of effective anomaly detection systems still presents numerous difficulties for the computer vision community. This includes the lack of cameras, bad weather, problems with night vision, etc. We have used several special methods that enhance the system's overall performance