Moving Trajectory Based Traffic Police Gesture Recognition Via Time Series Classification
Guanying Huang, Jun Jie Yang, Siyuan Jing · 2021 17th International Conference on Computational Intelligence and Security (CIS) · 2021
Gesture recognition plays an important role in the future human-computer interaction. This paper focuses on traffic police gesture recognition and introduces a novel method to handle the task by using time series classification technologies. Firstly, the proposed method captures the outline of the traffic police in video. Secondly, it selects several key points from the outline, and transforms the moving trajectory of these key points to a set of time series of which every value denotes the distance between the key point and the reference point. The proposed method can obtain accuracy 89% on a public data set, and the experiments prove that our method is effective.