Abstract:Gait-based pedestrian identification has important applications in intelligent surveillance. From anatomical viewpoint, the physical uniqueness of human gait is physiological discriminative of individuals. Therefore, in theory, like fingerprint and face, gait is used as a biometric for pedestrian identification. However, gait-based pedestrian identification still faces multiple challenges due to a vast diversity of walking conditions and complex acquisition environment during data collection. In this paper, through combining the nonlinear dimensionality reduction by using gait manifold and the temporal feature of gated recurrent unit (GRU) together, we propose a novel gait-based pedestrian identification framework. Firstly, we design a temporal enhancement module to construct a series of frame-by-frame gait trend energy images (ff-GTEIs), which represents spatiotemporal gait characteristics and does not reduce the number of samples. Secondly, a supervised locally linear embedding (LLE) dimensionality reduction scheme is proposed, which generates a low-dimensional gait manifold for each pedestrian and transforms all ff-GTEIs into corresponding gait manifold space. Thirdly, a new pedestrian identification network based on residual GRU is proposed, which is able to identify a person by comprehensively considering the similarity between its gait and corresponding gait manifold. Finally, a series of comparative experiments are carried out based on well-known gait datasets, the experimental results show that the proposed framework for pedestrian recognition in this paper exceeds most existing methods, and has achieved an average correct recognition rate 97.4% and 99.6% based on the open-accessed gait dataset CASIA Dataset B and OU-ISIR LP dataset, respectively.

GaitGANv2: Invariant Gait Feature Extraction Using Generative Adversarial Networks

Gait Recognition, Generative Adversarial Networks,cross View

Non-local Gait Feature Extraction and Human Identification

Multi-Task GANs for View-Specific Feature Learning in Gait Recognition

Robust Cross-View Gait Recognition with Evidence: A Discriminant Gait GAN (DiGGAN) Approach

Human Identification Based on Gait Manifold

Feature Separation GAN for Cross View Gait Recognition

A Novel Human Gait Recognition Method by Segmenting and Extracting the Region Variance Feature

Gait Recognition Using Multichannel Convolution Neural Networks

Human Gait Recognition Based on Frame-by-Frame Gait Energy Images and Convolutional Long Short-Term Memory

On Learning Disentangled Representations for Gait Recognition

Dense-View GEIs Set: View Space Covering for Gait Recognition based on Dense-View GAN

Human Gait Recognition Based on Self-Adaptive Hidden Markov Model

DyGait: Exploiting Dynamic Representations for High-performance Gait Recognition

Attacking Gait Recognition Systems Via Silhouette Guided GANs

DeepGait: A Learning Deep Convolutional Representation for View-Invariant Gait Recognition Using Joint Bayesian

Toward Complete-View and High-Level Pose-Based Gait Recognition

Frame-GAN: Increasing the Frame Rate of Gait Videos with Generative Adversarial Networks.

Make the Bag Disappear: Carrying Status-invariant Gait-based Human Age Estimation Using Parallel Generative Adversarial Networks

Adversarial learning-based skeleton synthesis with spatial-channel attention for robust gait recognition

Gait Energy Image-Based Human Attribute Recognition Using Two-Branch Deep Convolutional Neural Network