Abstract
In the modern era, the empirical analysis of countermeasures against crime has developed rapidly from the perspective of crime causes. The concept of punishment based on preventionism has been vigorously developed. We aimed to apply the principle of convolutional neural network (CNN) algorithm in deep learning to frame a crime prediction model in three-dimensional CNN (3D CNN) for representative driver behaviours offenses. Results indicated that the 3D CNN prediction model predicted road traffic offences through the detection of five driver offender behaviours with an accuracy rate of 96%. The prediction model we propose appears to hold great prospects for driver offender behaviour trending and to provide a reference for future studies on crime prevention.