3-Step Risk Assessment and Dissemination Framework with CCTV Video for Crime Prevention

Existing crime prediction systems have limitations in crime prevention, lack of monitoring personnel compared to the number of installed CCTV cameras, and reliance solely on conventional investigation data and experience for determining crime risk levels. To overcome these limitations, this paper introduces a novel machine learning-based CCTV video analysis and dissemination framework for crime prevention. The framework is developed for providing swift and accurate compressed data sets of CCTV videos through risk assessment and analysis algorithms for behavior and target recognition, time-based crime occurrence rates, and detection of abnormal behavior. In addition, it can resolve the blind spot issue of the traditional CCTV video systems by our breadth-first search mechanism.