Enhancing Gait Recognition with Attention-Based Spatial-Temporal Deep Learning: The GaitDeep Framework

• Published in: Computer science journal of Moldova Vol. 33; no. 2(98); pp. 188 - 218
• Main Authors: Mandlik, Sachin, Labade, Rekha, Chaudhari, Sachin
• Format: Journal Article
• Language: English
• Published: Vladimir Andrunachievici Institute of Mathematics and Computer Science 01.09.2025
• Subjects: biometric; deep learning; gait recognition; spatial-temporal refinement
• ISSN: 1561-4042; 2587-4330; 1561-4042; 2587-4330
• DOI: 10.56415/csjm.v33.10

Gait, an individual's unique walking style, serves as an effective biometric tool for surveillance. Unlike fingerprints or iris scans, gait is observable from a distance without the subject's awareness, making it ideal for security applications. CNNs struggle with video variability, affecting gait recognition. This study introduces GaitDeep, a spatial-temporal refinement using a deep dense network. It integrates attention-enhanced spatial extraction with a two-directional LSTM-based temporal module to prioritize key segments. Evaluated on the OU-ISIR, OU-MVLP, and CASIA-B datasets, GaitDeep achieves accuracies of 95.1%, 0.96%, and 98.10%, respectively, outperforming state-of-the-art methods and establishing a new benchmark for gait recognition.