[Paper Review] A Comprehensive Survey on Deep Gait Recognition: Algorithms, Datasets and Challenges

Chuanfu Shen, Shiqi Yu+3 more|arXiv (Cornell University)|Jun 28, 2022

Gait Recognition and Analysis30 citations

TL;DR

This paper provides an up-to-date survey of deep learning-based gait recognition, introducing a two-dimensional taxonomy (deep representations and architectures), reviewing datasets, evaluating performance across scenarios, and addressing privacy and security concerns.

ABSTRACT

Gait recognition aims to identify a person at a distance, serving as a promising solution for long-distance and less-cooperation pedestrian recognition. Recently, significant advancements in gait recognition have achieved inspiring success in many challenging scenarios by utilizing deep learning techniques. Against the backdrop that deep gait recognition has achieved almost perfect performance in laboratory datasets, much recent research has introduced new challenges for gait recognition, including robust deep representation modeling, in-the-wild gait recognition, and even recognition from new visual sensors such as infrared and depth cameras. Meanwhile, the increasing performance of gait recognition might also reveal concerns about biometrics security and privacy prevention for society. We provide a comprehensive survey on recent literature using deep learning and a discussion on the privacy and security of gait biometrics. This survey reviews the existing deep gait recognition methods through a novel view based on our proposed taxonomy. The proposed taxonomy differs from the conventional taxonomy of categorizing available gait recognition methods into the model- or appearance-based methods, while our taxonomic hierarchy considers deep gait recognition from two perspectives: deep representation learning and deep network architectures, illustrating the current approaches from both micro and macro levels. We also include up-to-date reviews of datasets and performance evaluations on diverse scenarios. Finally, we introduce privacy and security concerns on gait biometrics and discuss outstanding challenges and potential directions for future research.

Motivation & Objective

• Summarize the advances of deep gait recognition from a two-dimensional taxonomy (deep representations and neural architectures).
• Review datasets and benchmarking progress across cross-view, in-the-wild, cloth-changing, and 3D space scenarios.
• Discuss biometric security and privacy implications and outline outstanding challenges and future directions.

Proposed method

• Propose a novel two-dimensional taxonomy for deep gait recognition focused on deep representation learning and deep network architectures.
• Survey and categorize existing deep gait methods from micro (feature learning) and macro (architectures) perspectives.
• Provide up-to-date reviews of gait datasets and performance evaluations in four scenarios: cross-view, in-the-wild, cloth-changing, and 3D space.
• Discuss privacy and security concerns in gait biometrics and propose directions for future research.

Experimental results

Research questions

• RQ1 What are the current deep representation learning techniques used for gait recognition?
• RQ2 How do neural architectures (discriminative vs. generative) contribute to gait recognition performance?
• RQ3 What are the key datasets and how do methods perform across cross-view, in-the-wild, cloth-changing, and 3D-space scenarios?
• RQ4 What privacy and security issues threaten gait biometrics and what future directions address them?

Key findings

• The survey provides an up-to-date catalog of deep gait recognition methods, datasets, and benchmarks.
• A novel taxonomy is proposed that analyzes deep gait methods by deep representations learning and neural architectures, offering a macro/micro view beyond traditional model- vs. appearance-based classifications.
• Performance across scenarios is summarized, noting that deep models achieve strong results on datasets such as CASIA-B (e.g., high accuracy in appearance-changing settings) and GREW (over 70% rank-1 in outdoor, large-scale trials).
• The paper discusses biometric security and privacy concerns, highlighting challenges and potential future directions in gait biometrics.