A systematic review on CNN-YOLO techniques for face and weapon detection in crime prevention

The increase in criminal activities has led to an increased focus on crime forecasting and proactive measures using advanced technologies to detect crime activity before the crime has to be committed or suspected. A major challenge is to prevent such crimes and protect people from criminals with the help of deep learning technology with advanced neural networks, computer vision, classifiers, and methodologies such as you only look once, Faster RCNN, convolution neural networks, and GANs, which are employed to forecast crime activities while preventing criminal activities before they happen. Crime analysis revealed two key elements like facial recognition and weapon detection and these technologies are beneficial in identifying criminals and determining their exact situation. Currently, governments and private institutions use closed-circuit television (CCTV) based surveillance to prevent and monitor crime. These techniques provide high performance and accuracy in recognizing criminals. This review aims to highlight the current state of these technologies, their impact on crime analysis, and potential future efforts related to detection under difficult conditions, minimizing false alarms, and integrating detection models into real-time, scalable, and privacy-preserving surveillance systems developments to improve security and law enforcement practices.