In this paper, a method using viewpoint for person re-identification is proposed. The method divides a pedestrian image into local patches, extracts features from each patch and estimates angle each patch using person orientation. Then pedestrian imag...
In this paper, a method using viewpoint for person re-identification is proposed. The method divides a pedestrian image into local patches, extracts features from each patch and estimates angle each patch using person orientation. Then pedestrian images are matched by distance between local features and angle. The effectiveness of the proposed method is validated on a public dataset.