Anomaly Detection Methods in Surveillance Videos: A Survey
Lokesh Borawar, Ravinder Kaur · 2022
Surveillance system continuously generates massive amount of video data in the newest technological era, analysing these data is a tedious task for security specialists. With the popularization of surveillance monitoring system and the evolution of information technology, how to immediately and without human interaction detect unusual behaviours in surveillance footage is becoming more and more crucial for smart cities and public safety. Finding of abnormal footage physically in these massive video recordings is a laborious work, as they do not happen often in the real world. This clearly shows the necessity of automated anomaly detection, afterward that can detect crimes and aid investigations. The progress of anomaly detection has substantially benefited from deep learning, and much outstanding work has been published on this subject. This survey paper provides a comprehensive review of various anomaly detection and recognition methods. Researchers will get a better perspective of anomaly detection task with GAN approach, fine-tuned approach, and keyframes extraction plus shallow network approach and also their issues as well.