Abstract
For unsupervised domain adaption in person re-identification (Re-ID) tasks, the generally used label estimation approaches simply use the global features or the uniform part features. They often neglect the variations of samples having the same identity caused by occlusion, misalignment and uncontrollable camera settings. In this paper, we propose a discriminative learning network with target domain latent information (LatentDLN) to enhance the generalization ability of the Re-ID model. Specifically, to generate a discriminative and robust representation, two types of latent information in the samples from the target domain are explored by the multi-branch deep structure. First, the key points based valid region information is used to leverage the local and global cues in human body, and then a heuristic distance metric learning method based on the global features and the local features is proposed to effectively evaluate the similarity between different images. Second, the camera style transferred images are used as augmentation data to bridge the gap between different cameras in target domains. Moreover, the re-rank mechanism based on reciprocal neighbors is designed to improve the quality of the label estimation. Experimental results on Market-1501, DukeMTMC-ReID and MSMT17 datasets validate the significant effectiveness of the proposed LatentDLN for unsupervised Re-ID.
Highlights
The person re-identification (Re-ID) task aims at matching relative pedestrian gallery images with a given query pedestrian image, where the gallery images are taken by different cameras [1]
To further facilitate research into person Re-ID tasks, we propose a discriminative learning network with target domain latent information (LatentDLN) for unsupervised person Re-ID, which operates as follows
In LatentDLN, while exploring the latent information, to deal with the cases of feature missing caused by occlusion, low resolution, etc., we propose a heuristic distance metric learning method
Summary
The person re-identification (Re-ID) task aims at matching relative pedestrian gallery images with a given query pedestrian image, where the gallery images are taken by different cameras [1]. Re-ID has attracted attentions of many researchers due to its important role in the field of video surveillance, and a large number of methods have been proposed. Many existing Re-ID methods focus on supervised learning and have achieved satisfactory performance on labeled re-id benchmark datasets [2]–[7]. To learn robust and invariant representation features or to obtain matching similarity measurements, the supervised methods require sufficient manually tagged pairwise person images for each pair of camera views. The existing domain gap between different Re-ID datasets limits the scalability and availability of the supervised learning methods
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
