A Machine Learning and Deep Learning Integrated Model to Detect Criminal Activities
Kanij Fatema Arpa, Tanni Mittra, Tasnim Ferdous, Nusrat Jahan, Riffat Ara Khan Tayna, Mahamudul Hasan, Mohammad Rifat Ahmmad Rashid, Md Sawkat Ali · 2023
Crime is considered an intended execution of an act that is harmful to society and punishable under criminal law. Bangladesh as a developing country has experienced severe crime-related issues in recent years due to the constantly growing criminal activities. The law enforcement sector of Bangladesh is dedicated to implementing the law and reducing fear of offense to make Bangladesh a secure place to live. Despite this, the traditional ways of detecting crimes in Bangladesh are currently slow and less effective because they are not getting enough technological support. As a result, most of the time the crimes get reported after the occurrence. So to attain efficiency in policing activities it has become significant to develop such a system that can notify the police department before the occurrence of a crime. In our proposed model, we develop a system that can prevent crime by detecting and notifying real-time criminal activities beforehand. To carry out this concept, we build a central system consisting of three modules i.e. criminal activity identification module, weapon detection module, and criminal detection module. Each of the modules is centrally connected and can give an alert when any type of abnormality occurs. We use Deep Learning and Machine Learning algorithms to perform criminal activity detection and face recognition of criminals. For each of the modules, we collect and prepare our own data set. We perform various experiments to investigate the effectiveness of the method. The evaluation result shows that our proposed method performs impressively in identifying criminal activities and criminals.