Summary of Adaptive Intra-class Variation Contrastive Learning For Unsupervised Person Re-identification, by Lingzhi Liu et al.
Adaptive Intra-Class Variation Contrastive Learning for Unsupervised Person Re-Identification
by Lingzhi Liu, Haiyang Zhang, Chengwei Tang, Tiantian Zhang
First submitted to arxiv on: 6 Apr 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 research paper proposes an adaptive intra-class variation contrastive learning algorithm for unsupervised person Re-ID, called AdaInCV. The method addresses limitations in existing approaches by introducing two new strategies: Adaptive Sample Mining (AdaSaM) and Adaptive Outlier Filter (AdaOF). AdaSaM dynamically refines the memory by creating more reliable clusters, while AdaOF identifies and filters out valuable outliers as negative samples. By considering intra-class variations after clustering, the algorithm evaluates the learning ability of the model for each class, selecting appropriate samples during training. The proposed approach achieves improved generalization abilities and addresses issues with false-positive samples. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The researchers developed a new way to improve person Re-ID without using labeled data. They created an algorithm that adjusts its sample selection based on how well it clusters people. This helps the model avoid mistakes by focusing on the most challenging examples. Additionally, they designed a filter to remove noise in the data, which improves overall performance. |
Keywords
» Artificial intelligence » Clustering » Generalization » Unsupervised
