Person Re-identification (ReID) is a critical computer vision task that aimsto match the same person in images or video sequences . Resolution is acrucial factor in person ReID when the cameras are at different distances from the person or the camera’s model are different from each other . In this paper, we propose a novel two-stream network with a lightweight resolution associationReID feature transformation (RAFT) module and a self-weighted attention (SWA)ReID module to evaluate features under different resolutions . RAFT transformsthe low resolution features to corresponding high resolution features . SWAevaluates both features to get weight factors for the person . Both modules jointly trained to get a resolution invariant representation .

Author(s) : Ziyue Zhang, Shuai Jiang, Congzhentao Huang, Richard Yi Da Xu

Links : PDF - Abstract

Code :
Coursera

Keywords : person - resolution - reid - features - transformation -

Leave a Reply

Your email address will not be published. Required fields are marked *