A review on Gait Recognition, Classification methods and Databases used

Abstract

Vulnerability to human authentication is a major problem in today's digital society in real-time settings such as airports, hospitals, metro stations, etc. Video surveillance security solutions have grown in popularity as a result of this problem. Due to its subtle and imperceptible character, behavioural biometric feature gait has become a viable candidate for a surveillance monitoring system in recent years. One benefit of even more human gait is that it may be observed in low-resolution footage and from a distance. Lastly, it is challenging to mimic gait characteristics. In this paper we cover gait recognition system and its types. The paper delves further into different classification methods which are mainly used for gait recognition. Additionally, the paper covers description of databases that can be used for research, divided into two groups: sensor-based and Vision Based. We look closely at the variables that influence gait recognition; recent work was done to address these variables. Furthermore, after reviewing the most recent research in the field, this paper suggests CNN based architecture which has recognition the person with high accuracy as compare to other discussed methods. Finally, we also provided a brief overview of the suggested workflow.