GaitGAN: Invariant Gait Feature Extraction Using Generative Adversarial Networks

Yu, Shiqi; Chen, Haifeng; García-Reyes, Edel; Poh, Norman

doi:10.1109/cvprw.2017.80

Cited by 167 publications

(94 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In addition to the commonly used CNN and LSTM network models, in order to solve the effects of perspective, weight, clothing and other issues on the extraction of gait intrinsic features, [71] designed a gait feature extractor based on the generative adversarial nets. Sometimes, gait datasets may contain many subjects while each of these subjects only has a small amount of data, which may not support the training of deep models.…”

Section: B Deep Learning For Gait Recognitionmentioning

confidence: 99%

Deep Learning-Based Gait Recognition Using Smartphones in the Wild

Zou

Wang

et al. 2020

IEEE Trans.Inform.Forensic Secur.

221

141

View full text Add to dashboard Cite

Comparing with other biometrics, gait has advantages of being unobtrusive and difficult to conceal. Inertial sensors such as accelerometer and gyroscope are often used to capture gait dynamics. Nowadays, these inertial sensors have commonly been integrated in smartphones and widely used by average person, which makes it very convenient and inexpensive to collect gait data. In this paper, we study gait recognition using smartphones in the wild. Unlike traditional methods that often require the person to walk along a specified road and/or at a normal walking speed, the proposed method collects inertial gait data under a condition of unconstraint without knowing when, where, and how the user walks. To obtain a high performance of person identification and authentication, deep-learning techniques are presented to learn and model the gait biometrics from the walking data. Specifically, a hybrid deep neural network is proposed for robust gait feature representation, where features in the space domain and in the time domain are successively abstracted by a convolutional neural network and a recurrent neural network. In the experiments, two datasets collected by smartphones on a total of 118 subjects are used for evaluations. Experiments show that the proposed method achieves over 93.5% and 93.7% accuracy in person identification and authentication, respectively.Index Terms-Gait recognition, inertial sensor, person identification, convolutional neural network, recurrent neural network.

show abstract

Section: B Deep Learning For Gait Recognitionmentioning

confidence: 99%

Deep Learning-Based Gait Recognition Using Smartphones in the Wild

Zou

Wang

et al. 2020

IEEE Trans.Inform.Forensic Secur.

221

141

View full text Add to dashboard Cite

show abstract

“…We fix probe view angle and average accuracies of different gallery view. Comparing to GaitGAN [13], our method get better performance on each view angle. The results show that besides view variation, our method can deal with different walking conditions.…”

Section: Experimental Results On Casia-b Datasetmentioning

confidence: 87%

“…Zhang et al [12] developed a Siamese neural network based gait recognition framework to automatically extract robust and discriminative gait features for human identification. Yu et al [13] proposed a method named as GaitGAN which is based on generative adversarial networks (GAN) [14]. In the proposed method, a GAN model is taken as a regressor to generate invariant gait images that is side view images with normal clothing and without carrying bags.…”

Section: Introductionmentioning

confidence: 99%

Gait recognition, generative adversarial networks,cross view

Zhang¹,

Yin²,

Zhou³

et al. 2019

2019 Scientific Conference on Network, Power Systems and Computing (NPSC 2019)

View full text Add to dashboard Cite

The performance of gait recognition can be obviously affected by view angle variation. In this paper, we present a new method which uses a view transformation generative adversarial networks (GAN) to improve performance of dealing with cross-view gait recognition problem. Our proposed method firstly trains a convolutional neural network (CNN) using gait energy image (GEI) for recognition. Then, a GAN model is taken as a generator to transform gait images with variety view angle to unique side view images. In order to preserve the identification information of generated images, the generated images are input into the fixed pre-trained CNN and recognition loss is used to update generator. Finally, we combine the distance matrix of original and generated image and get final recognition results. We conduct experiments to demonstrate the improvement of adding GAN branch on three popular gait dataset. Experimental results show that our method can achieve state-of-the-art performance.

show abstract

“…McLaughlin 等 [37] 则提出了一种结合了卷积网络和循环网络的新网络训练步态视频嵌入, 并且用于步态重识别任务. 另外, Yu 等 [38] 还探索了用生成对抗网络提取步态特征的方法.…”

Section: 研究背景unclassified