Summary of Exploring Homogeneous and Heterogeneous Consistent Label Associations For Unsupervised Visible-infrared Person Reid, by Lingfeng He et al.
Exploring Homogeneous and Heterogeneous Consistent Label Associations for Unsupervised Visible-Infrared Person ReID
by Lingfeng He, De Cheng, Nannan Wang, Xinbo Gao
First submitted to arxiv on: 1 Feb 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper presents a novel approach to unsupervised visible-infrared person re-identification, dubbed USL-VI-ReID. The goal is to retrieve pedestrian images of the same identity from different modalities without annotations. Existing methods focus on establishing cross-modality pseudo-label associations, but neglect maintaining instance-level consistency between feature and pseudo-label spaces. To address this, the authors introduce a Modality-Unified Label Transfer (MULT) module that accounts for both homogeneous and heterogeneous fine-grained instance-level structures, generating high-quality cross-modality label associations. The proposed MULT ensures alignment across modalities while upholding structural consistency within intra-modality. Additionally, an Online Cross-memory Label Refinement (OCLR) module is introduced to mitigate noisy pseudo-labels and align different modalities. The authors demonstrate the superiority of their approach over existing state-of-the-art methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper helps us better recognize people from different camera angles or light conditions without needing human labels. Right now, we can’t do this very well because our computers don’t understand how to connect pictures taken in different ways (like daytime and nighttime). To fix this, the authors created a new way to match pictures that takes into account the little details that make one picture look like another (like clothes or height). This helps our computers be better at recognizing people from different angles. The authors also showed that their approach is more accurate than other methods. |
Keywords
» Artificial intelligence » Alignment » Unsupervised
