Manuscript received November 6, 2023; revised January 15, 2024; accepted January 26, 2024; published June 6, 2024

Abstract—Identifying individuals based on their gait is a crucial aspect of biometric authentication. It is complicated by several factors, such as altering one’s walking posture, donning a coat, and wearing high heels. With the advent of artificial intelligence, deep learning, in particular, has made significant strides in this area. The conditional Generative Adversarial Network (cGAN), together with hybrid Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNNs), are used in this research to create images using a novel technique. The framework comprises three parts. The first involves extracting silhouettes, necessitating computing the gait cycle and energy. The technique of creating images using discriminator models and cGANs is the second part. Image classification using hybrid LSTM and CNN networks is the third step. Experiments were conducted to assess our approach using the CASIA database, a publicly available gait recognition dataset. Our proposed approach achieved a high classification accuracy of 97.11%. Our results outperform state-of-the-art techniques, especially when it comes to carrying bags and donning coats.

Keywords—gait recognition, Generative Adversarial Networks (GANs), Long Short-Term Memory (LSTM), deep learning

Cite: Entesar T. Burges, Zakariya A. Oraibi, and Ali Wali, "Gait Recognition Using Hybrid LSTM-CNN Deep Neural Networks," Journal of Image and Graphics, Vol. 12, No. 2, pp. 168-175, 2024.

Copyright © 2024 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC BY-NC-ND 4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.