Person Matching Method of Gait Information with 2D Pose Estimation

Abstract

Recently, research and development for matching individuals using various biological information such as face images and veins has been advanced. In the case of a face image, for example, a conscious action is required in order to satisfy a specific condition such as a face facing the camera. Therefore, in a situation where a person is not conscious of collation, such as in a security camera image, collation becomes difficult. In this paper, was developed a technique to verify the same person from pedestrians with different images by using two-dimensional posture estimation information. In addition, the results of comparison with the proposed method are reported.