Center-Level Verification Model for Person Re-identification
In past years,convolutional neural network is increasingly used in person re-identification due to its promising performance.Especially,the siamese network has been widely used with the combination of verification loss and identification loss.However,the loss functions are based on the individual samples,which cannot represent the distribution of the identity in the scenario of deep learning.In this paper,we introduce a novel center-level verification(CLEVER)model for the siamese network,which simply represents the distribution as a center and calculates the loss based on the center.To simultaneously consider both intra-class and inter-class variation,we propose an intra-center submodel and an inter-center submodel respectively.The loss of CLEVER model,combined with identification loss and verification loss,is used to train the deep network,which gets state-of-the-art results on CUHK03,CUHK01 and VIPeR datasets.
Center-level Intra-class variation Inter-class distance
Ruochen Zheng Yang Chen Changqian Yu Chuchu Han Changxin Gao Nong Sang
Key Laboratory of Ministry of Education for Image Processing and Intelligent Control,School of Automation,Huazhong University of Science and Technology,Wuhan 430074,China
国际会议
广州
英文
98-107
2018-11-23（万方平台首次上网日期，不代表论文的发表时间）