Dual-Resolution Fusion Modeling for Unsupervised Cross-Resolution Person Re-Identification

Published: 20 Jul 2024, Last Modified: 06 Aug 2024MM2024 OralEveryoneRevisionsBibTeXCC BY 4.0
Abstract: Cross-resolution person re-identification (CR-ReID) aims to match images of the same person with different resolutions in different scenarios. Existing CR-ReID methods achieve promising performance by relying on large-scale manually annotated identity labels. However, acquiring manual labels requires considerable human effort, greatly limiting the flexibility of existing CR-ReID methods. To address this issue, we propose a dual-resolution fusion modeling (DRFM) framework to tackle the CR-ReID problem in an unsupervised manner. Firstly, we design a cross-resolution pseudo-label generation (CPG) method, which initially clusters high-resolution images and then obtains reliable identity pseudo-labels by fusing class vectors in both resolution spaces. Subsequently, we develop a cross-resolution feature fusion (CRFF) module to fuse features from both high-resolution and low-resolution spaces. The fusion features have the potential to serve as a new form of resolution-invariant features. Finally, we introduce cross-resolution contrastive loss and probability sharpening loss in DRFM to facilitate resolution-invariant learning and effectively utilize ambiguous samples for optimization. Experimental results on multiple CR-ReID datasets demonstrate that the proposed DRFM not only outperforms existing unsupervised methods but also approaches the performance of early supervised methods.
Primary Subject Area: [Experience] Multimedia Applications
Relevance To Conference: Person re-identification represents a fundamental challenge in computer vision, particularly for applications such as surveillance, security, and human behavior analysis. Existing cross-resolution person re-identification (CR-ReID) methods show promising effectiveness, but they heavily rely on manual annotation efforts to provide large-scale identity labels. However, the manual labeling process requires significant human involvement, limiting the flexibility of current CR-ReID techniques. Given the absence of unsupervised CR-ReID methods, this work proposes a dual-resolution fusion modeling (DRFM) framework to address the CR-ReID problem in an unsupervised manner. In DRFM, we introduce a cross-resolution pseudo-label generation method to obtain reliable pseudo-labels and employ cross-resolution feature fusion and multiple optimization methods to facilitate resolution-invariant feature learning. Extensive experiments conducted on three datasets demonstrate the effectiveness and superiority of DRFM. In conclusion, our work advances the progress in multimedia and multimodal processing, providing strong support for various real-world applications reliant on accurate person re-identification across different environments and conditions.
Submission Number: 2376
Loading