Exploring Artificial Intelligence-Driven Gait Analysis for Suspect Identification in Forensic Video Footage

Abstract

In our digitally connected world, surveillance cameras record massive amounts of visual data on a daily basis. This paper takes a close look at how artificial intelligence, particularly methods centered on gait analysis, could be harnessed for criminal investigations when facial recognition proves insufficient. Our primary focus is to assess how a person’s distinctive walking pattern can aid in identifying a potential suspect in video footage used for forensic purposes. We provide an extensive background on both conventional and emerging approaches to gait recognition, then outline a unified methodology that incorporates denoising, feature extraction, and neural network-based classification. We also present experimental evaluations that illuminate the reliability and shortcomings of gait-based investigative techniques, with specific attention to real-world complications such as altered viewpoints, clothing variations, and partial obstructions. Overall, the aim here is to furnish a comprehensive picture of AI-powered gait analysis and to underscore its potential to augment existing forensic methods while emphasizing the practical steps required to ensure its efficacy and reliability in real investigations.