Cross-Resolution Person Re-identification with Deep Antithetical Learning

Abstract

Images with different resolutions are ubiquitous in public person re-identification (ReID) datasets and real-world scenes, it is thus crucial for a person ReID model to handle the image resolution variations for improving its generalization ability. However, most existing person ReID methods pay little attention to this resolution discrepancy problem. One paradigm to deal with this problem is to use some complicated methods for mapping all images into an artificial image space, which however will disrupt the natural image distribution and requires heavy image preprocessing. In this paper, we analyze the deficiencies of several widely-used objective functions handling image resolution discrepancies and propose a new framework called deep antithetical learning that directly learns from the natural image space rather than creating an arbitrary one. We first quantify and categorize original training images according to their resolutions. Then we create an antithetical training set and make sure that original training images have counterparts with antithetical resolutions in this new set. At last, a novel Contrastive Center Loss(CCL) is proposed to learn from images with different resolutions without being interfered by their resolution discrepancies. Extensive experimental analyses and evaluations indicate that the proposed framework, even using a vanilla deep ReID network, exhibits remarkable performance improvements. Without bells and whistles, our approach outperforms previous state-of-the-art methods by a large margin.

Part-Based Enhanced Super Resolution Network for Low-Resolution Person Re-Identification

“…Another popular solution is to elevate LR images to a uniform HR [10], [11], [30], followed by identification tasks. In work [28], Zhuang et al cascaded multiple SRGANs in series and plug-in a REID network to get the consistent HR images. Mao et al [31] proposed a similar model, which learned similar resolution-invariant image representations and was able to recover the missing details in LR input images that is beneficial to improve person re-ID performance.…”

SR-DSFF and FENet-ReID: A Two-Stage Approach for Cross Resolution Person Re-Identification

“…We believe that this will bring hidden dangers to subsequent Re-ID tasks. Although Zhuang et al [ 21 ] proposed CAD-NET to jointly learn the feature maps of the SR images and the LR images to alleviate the loss of feature details; however, there are still significant problems in directly fusing feature maps from images of different resolutions. Furthermore, most researchers use deep neural networks to capture low-level details by extracting local features [ 22 ] of images, which are likely to bring semantic ambiguity.…”