Determining the Location of a Concealed Handgun on the Human Body using Marker-Less Gait Analysis and Machine Learning

There has been a consistent rise in crimes involving handguns over the years globally. A majority of these handguns are concealed on the hip area to the location where they will be used to commit these crimes. The possible concealment locations on the hip could be the left, right, crouch and back of the hip. A Handgun concealed on the hip has been found to result in a disruption in gait. Marker-less gait analysis approach on video using machine learning techniques have been found to achieve exemplary detection rates especially for handguns concealed on the right hip. This study aims to extend this approach by developing machine learning models with the ability to determine the exact location on the hip where a handgun is concealed. The K-Nearest neighbor and Random forest algorithms presented the best performance with over 90% accuracy and precision of determining the location of the handgun and less than 4% false alarm rate. This performance demonstrates the viability of the approach. The approach can be implemented in CCTV networks and will enable security personnel to have complete information of the location of the handgun when approaching potential suspects.