Enhancing Digital Forensics with Deep Learning: Applications of CNNs and YOLOv5 in Weapon Detection and Image Forgery Analysis

Security is an essential problem in all domains. The crime rates are increased at crowded events or suspected isolated regions. Computer vision has significant uses in detecting and monitoring abnormalities to address diverse issues. Video surveillance systems are increasingly necessary for safeguarding the safety, security, and personal belongings. The ability of these systems to identify and understand scenes and unusual occurrences is crucial for effective intelligence monitoring. The primary objective of this study was to analyze surveillance films to identify weapons and detect any unusual behaviors or actions. This study applied the advanced cutting-edge framework YOLOv5 to examine and identify abnormalities like weapon recognition and criminal behavior in the public surveillance video dataset. The proposed approach implementation accomplishes an (mAP) mean average precision of 96.1%, outperforming state-of-the- art methods in terms of accuracy and efficiency for recognizing weapons and localization of criminal behavior in challenging surveillance datasets.