A Cross-Scene Person Re-Identification Through Gait Recognition Using Silhouette-Based Deep Learning Model

Person Re-Identification (Re-ID) through gait analysis is gaining attention as a powerful and practical method for recognizing people across different camera views without depending on facial features or what they are wearing.Since the way a person walks are distinctive and tends to stay consistent even with changes in clothing, lighting, or camera angles, gait offers a reliable biometric for long-distance surveillance and security.In this study, a simplified and effective framework for person Re-ID that relies on analyzing how people walk in video footage is presented.The system works by first extracting a person's silhouette and then using Deep Learning (DL) to understand both how their body looks and how it moves over time.To do this, Convolutional Neural Networks (CNNs) to capture visual details with Recurrent Neural Networks (RNNs) to track motion across frames are combined.This combination helps the system better recognize and tell individuals apart based on their unique walking patterns.The system is tested on publicly available gait datasets and found that it performs exceptionally well, even under different conditions.The system also includes a detection component to automatically identify and track people across different scenes before applying the gait recognition process.Our experimental results show that the method is highly robust in real-world situations, making it a promising tool for applications like surveillance, access control, and forensic analysis.This research moves forward the development of nonintrusive, reliable technologies for person Re-ID using gait.