Anomaly Detection in Surveillance Camera

In high security locations, anomaly detection in surveillance systems is essential for maintaining public safety. This work proposes an object detection approach utilizing model renowned for its real-time performance for anomaly detection. The system perceives objects and trails actions by using YOLOv5 where it notes down the odd behaviors as perhaps in security video where it abnormal. A dataset which contains of several surveillance synopsis is used to train and evaluate the model. YOLOv5 has such an object detection capability with a custom made revelation framework which grants the system to differentiate between abnormal and legal activities in real-time. An emphasis on prudent processing speed is done to provide scalability in practical applications and the high authenticity abnormality detection is determined by experimental findings.