The Need for marker-less computer vision techniques for human gait analysis on video surveillance to detect concealed firearms

Abstract

Crimes involving the use of firearms have been on the increase in the past few years. One of the measures adopted to prevent these crimes is the use of CCTV operators at video surveillance centers to detect persons carrying concealed firearms on their bodies by monitoring their behavior. This paper has found that this technique has challenges associated with human weaknesses and errors. A review of the current attempts to automate video surveillance for concealed firearm detection has found that they have the limitation that the techniques can only be employed on stationary and cooperative persons. This makes them inappropriate for real-life surveillance. This paper highlights the need for automated video surveillance solutions that can detect persons carrying concealed firearms when they are not stationary and aware of the scanning process. We further explore automated behavioral analysis and specifically gait analysis as a possible technique for concealed firearm detection on video surveillance. Lastly, the paper highlights the possibility and viability of human gait analysis using marker-less computer vision techniques for detecting persons carrying firearms on their waist line.