Deep learning based Person Re-Identification method has recently attracted a lot of research interests. But it requires a great number of high quality images when sometimes can not be satisfied. In this paper, we transform the RGB color space to our color space based on human’s visual experience and calculate histograms of low quality images. Then we use 4 distance metric methods to calculate histogram similarities between each pair of images and match. Moreover, we use semantic information into our matching method. The experimental results show that semantic information is able to improve the accuracy of color histogram based re-id method.
I22-40
助成情報
助成機関名
日本学術振興会
Japan Society for the Promotion of Science (JSPS)
Mendeley出力
