Visible-infrared person re-identification has attracted extensive attention from the community due to its potential great application prospects in video surveillance. There are huge modality discrepancies between visible and infrared images caused by different imaging mechanisms. Existing studies alleviate modality discrepancies by aligning modality distribution or extracting modality-shared features on the original image. However, they ignore a key solution, i.e., converting visible images to gray images directly, which is efficient and effective to reduce modality discrepancies. In this paper, we transform the cross-modality person re-identification task from visible-infrared images to gray-infrared images, which is named as the minimal modality discrepancy. In addition, we propose a pyramid feature integration network (PFINet) which mines the discriminative refined features of pedestrian images and fuses high-level and semantically strong features to build a robust pedestrian representation. Specifically, PFINet first performs the feature extraction from concrete to abstract and the top-down semantic transfer to obtain multi-scale feature maps. Second, the multi-scale feature maps are inputted to the discriminative-region response module to emphasize the identity-discriminative regions by the spatial attention mechanism. Finally, the pedestrian representation is obtained by the feature integration. Extensive experiments demonstrate the effectiveness of PFINet which achieves the rank-1 accuracy of 81.95% and mAP of 74.49% on the multi-all evaluation mode of the SYSU-MM01 dataset.

In this paper, we present a practical system to automatically suggest the most pairing clothing items, given the reference clothing (upper-body or low-body). This has been a challenge, due to clothes having a variety of categories. Clothing is one of the most informative cues for human appearance. In our daily life, people need to wear properly and beautifully to show their confidence, politeness and social status in various occasions. However, it is not easy to decide what to wear and how to coordinate their own clothes. To address this problem, we propose a recommendation approach that includes clothing region detection, clothing pair recommendation and distance fusion. Clothing region detection based on Faster R-CNN is used to detect clothing region. Clothing pair recommendation consists of a quadruple network architecture, where one dual network of the architecture adopts Siamese convolution neural network architecture. Training examples are pairs of upper-body and low-body clothing items that are either compatible or incompatible. The other dual network is used to learn clothing style features of the input image. This framework is designed to learn a feature transformation from the images of clothing items into two latent spaces, which we call them compatible space and style space respectively. After training the two dual networks, we use a distance fusion method to fuse the features extracted from the compatible and style dual networks. To acquire an optimized model and verify our proposed method, we expand an existing large clothing dataset WoG (Weather-to-Garment), and the resulted dataset is called “How to Wear Beautifully” (H2WB). Experiments on the H2WB dataset demonstrate that our approach is effective with clothing region detection and clothing pair recommendation as well as distance fusion.