Multi‐scale feature fusion network for person re‐identification. Issue 17 (25th February 2021)

• Record Type: Journal Article
• Authors: Wang, Yongjie; Zhang, Wei; Liu, Yanyan
• Is Part Of: IET image processing. Volume 14:Issue 17(2020)
• Page Start: 4614
• Page End: 4620
• Publication Date: 2021-02-25
• Subjects: feature extraction -- object recognition -- image classification -- pose estimation -- learning (artificial intelligence) -- image fusion -- convolutional neural nets; Res5 -- average pooling layer -- max‐pooling layer -- triplet loss function -- Res4 -- stride pooling layer -- softmax loss function -- averaged triplet losses -- averaged softmax losses -- multiscale feature fusion network -- effective person re‐identification -- convolutional neural network -- body analysis -- mixed pooling strategy -- mixed loss function -- pyramid pooling structure -- visual clues; Image processing -- Periodicals; 621.36705
• DOI: 10.1049/iet-ipr.2020.0008
• Languages: English
• ISSNs: 1751-9659

Abstract

Abstract : Recently, it is becoming a challenging work for person re‐identification due to the problems of occlusion, blurring and posture. The key of effective person re‐identification is to capture sufficient detailed features of a person's appearance in images. Different from previous methods, our method mainly focuses on fusing different visual clues only depending on the features of different levels and scales without additional assistance. The major contributions of our paper are the mixed pooling strategy with different kernels and the mixed loss function. Firstly, we adopt ResNet50 as our backbone. We have slightly modified the backbone, which does not use the down‐sampling operation at the beginning of stage 4. Inspired by pyramid pooling structure, we pass the outputs of Res4 and Res5 through the average pooling layer and max pooling layer with different kernels and strides separately. Secondly, we combine the averaged triplet losses and the averaged softmax losses as the final loss of the whole network. Extensive experiments on three datasets (CUHK3, Market1501, DukeMTMC‐reID) show that compared with many state‐of‐the‐art methods in recent years, our model achieve higher accuracy.