Intelligent Criminal Identification System For Semi-Regulated Environments

Abstract

Nowadays, algorithms and technologies for criminal identification systems are imperative for the fight against crime in every environment. Thus, closed-circuit television cameras are now being incorporated into physical security mechanisms to identify suspects and criminals at crime scenes. However, crime investigators often spend a great deal of time unravelling the identities of criminals in lengthy video footage, especially in a semi-regulated environment where both known and unknown faces are always present. However, the research that combined a Single Short MultiBox Detector (SSD) and Local Binary Patterns Histograms (LBPH) for criminal identification in a semi-regulated environment is yet to be explored. Hence, this study developed a system that proactively identifies known criminals or documents the presence of non-criminals for reactive investigation by combining SSD with LBPH. To achieve this, SSD was employed to detect faces, while the LBPH algorithm for face recognition. Also, the system sends security alerts via email and short message service once a criminal is identified in the environment. Based on the faces captured by the identification system, SSD and LBPH achieved a precision of 95% and 90% respectively. With this achievement, the system will not only reduce the time taken to identify criminals in a semi-regulated environment, but also will improve the chances of reporting the presence of confirmed criminals.