Human identification at a distance via gait recognition

Human identification at a distance via gait recognition

Human identification at a distance is a very challenging task, which has long been a popular research topic in the field of computer vision. The gait sequences of different people can be very distinctive, which makes gait an important body characteristic that can be used for human identification. However, to the best of our knowledge, there is no tutorial concerning this important area currently.

In this tutorial, we will first introduce the brief history of gait-based human identification and list out the challenges that lie in this field, such as cross-view and cross walking condition gait recognition. Then we will share a comprehensive survey on the different modules of a gait-based human identification system. Specifically, we will summarize both the traditional approaches and the advanced deep learning based approaches for gait-based human identification. In particular, such novel deep learning models can achieve an average accuracy of 98% under identical view conditions and 91% for cross-view scenarios in the database with more than 4000 people, which are much better than the previously reported results. Afterwards, we discuss the applications of gait recognition at a distance in different kinds of visual tasks. Finally, we provide the suggestions of employing gait recognition in practice and indicate potential directions of this area for future work.