GLOBAL TIMES: Chinese cities pilot gait recognition system

Beijing-based firm WATRIX reveals a tracking system that can mobilize tens of thousands of real-time cameras to recognize a person’s gait and help police nab suspects. Its pilot system is being used in some Chinese cities

Some Chinese cities are piloting a tracking system that can mobilize tens of thousands of real-time cameras to recognize a person’s gait and help police nab suspects, the firm that develops the technology revealed to the Global Times on Thursday.

Developed by the Beijing-based WATRIX, a company incubated by Institute of Automation under the Chinese Academy of Sciences, the system can locate and track targeted suspects by monitoring their postures from ten thousands of either real-time or offline videos.

By analyzing a person’s posture, the system can precisely and automatically track a person from videos and issue an alarm at any time if it spots someone doing something suspicious or illegal. The system is designed to relieve public security authorities from manually sifting through vast amounts of video to identify a suspect.

Traditional monitoring systems store and playback video but it is often unable to quickly and accurately identify, locate and find suspects due to poor video quality or if a suspect wears a disguise.

WATRIX told the Global Times that a pilot system is being used in Central China’s Hubei and South China’s Guangdong provinces, and Shanghai.

The system can be used in public places including airports, bus stations, and schools or used to detect intrusions into key infrastructure facilities such as nuclear power stations and oil refineries, the team said.

Every person’s posture is unique, like a fingerprint, and gait recognition technology is capable of secretly identifying targets from any angle, even if their face is covered and at night, Huang Yongzhen, CEO of WATRIX, told the Global Times earlier.

The technology will supplement China’s campaign to make Chinese cities and rural areas safer, which includes the projects Xueliang, or Sharp Eyes and Safe City, a professor at the National Defense University of the People’s Liberation Army in Beijing who requested anonymity told the Global Times on Thursday.

The Sharp Eyes project is a surveillance network in rural areas using artificial intelligence, facial recognition and big data. The Safe City project is an urban security network that uses a city’s surveillance camera network to detect crimes and traffic flow.

Using AI, facial recognition, big data and gait recognition technology, public security departments will be able to better solve crimes and reduce the crime rate, the professor noted.

Addressing privacy concerns, the team said the data captured by the system is only accessible to authorized users, and the company cannot access to the system.

Gait recognition technology can also be used to design smart home furniture, according to the team.

文章来源:http://www.globaltimes.cn/content/1156769.shtml

Human Identification at a Distance by Gait and Face Analysis —— A tutorial for ECCV2018

The outline of the half-day tutorial:

i. Introduction and overview of the tutorial: Motivations, challenges, available gait and face datasets. (15~20 minutes)

ii. A comprehensive survey on the whole pipeline of gait- and face-based human identification. (50~60 minutes)

1. Traditional approaches for gait- and face- based human identification at a distance
a) Image representation.
b) Feature dimensionality reduction.
c) Classification.

2. Advanced deep learning approaches for gait- and face- based human identification at a distance.
a) The network architecture design for gait and face recognition
b) The influencing factors to the performance such as input features, input resolution, temporal information, data augmentation, etc.
c) State-of-the-art gait and face recognition results on common benchmarks

iii. Applications of gait and face recognition in different kinds of visual tasks. (30~40 minutes)

iv. Suggestions in practice and discussion on potential directions. (10~15 minutes)

v. Experience the newest gait recognition system. (15~30 minutes)

vi. Open questions and discussion. (15~30minutes)

A Comprehensive Study on Cross-View Gait Based Human Identification with Deep CNNs

Introduction

This paper studies an approach to gait based human identification via similarity learning by deep convolutional neural networks (CNNs). With a pretty small group of labeled multi-view human walking videos, we can train deep networks to recognize the most discriminative changes of gait patterns which suggest the change of human identity. To the best of our knowledge, this is the first work based on deep CNNs for gait recognition in the literature. Here, we provide an extensive empirical evaluation in terms of various scenarios, namely, cross-view and cross-walking-condition, with different preprocessing approaches and network architectures. The method is first evaluated on the challenging CASIA-B dataset in terms of cross-view gait recognition. Experimental results show that it outperforms the previous state-of-the-art methods by a significant margin. In particular, our method shows advantages when the cross-view angle is large, i.e., no less than 36 degree. And the average recognition rate can reach 94 percent, much better than the previous best result (less than 65 percent). The method is further evaluated on the OU-ISIR gait dataset to test its generalization ability to larger data. OU-ISIR is currently the largest dataset available in the literature for gait recognition, with 4,007 subjects. On this dataset, the average accuracy of our method under identical view conditions is above 98 percent, and the one for cross-view scenarios is above 91 percent. Finally, the method also performs the best on the USF gait dataset, whose gait sequences are imaged in a real outdoor scene. These results show great potential of this method for practical applications.

Keywords

Deep learning, CNN, human identification, gait, cross-view

Authors

ZifengWu, Yongzhen Huang, Liang Wang, Xiaogang Wang and Tieniu Tan