Summary of Contrastive Multiple Instance Learning For Weakly Supervised Person Reid, by Jacob Tyo and Zachary C. Lipton
Contrastive Multiple Instance Learning for Weakly Supervised Person ReID
by Jacob Tyo, Zachary C. Lipton
First submitted to arxiv on: 12 Feb 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG)
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 Medium Difficulty Summary: This paper addresses the challenge of acquiring large-scale, precisely labeled datasets for person re-identification (ReID). Weakly supervised ReID has shown promise, but its performance lags behind fully supervised methods. To bridge this gap, the authors introduce Contrastive Multiple Instance Learning (CMIL), a novel framework that leverages contrastive losses without requiring pseudo labels or multiple models. CMIL outperforms state-of-the-art methods on three datasets, including SYSU-30k, WL-market1501, and WL-MUDD. The latter is a newly introduced dataset featuring naturally occurring weak labels from real-world applications. This framework’s ability to match state-of-the-art performance with fewer assumptions makes it an attractive solution for large-scale ReID tasks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Low Difficulty Summary: Imagine trying to find someone in a crowded place, like a mall or airport, without knowing what they look like. That’s kind of like the challenge faced by computer systems that need to recognize people from photos. One way to make this task easier is to use “weakly supervised” learning, which means training the system using imperfect information. Researchers have developed a new approach called Contrastive Multiple Instance Learning (CMIL) that can do this without needing lots of labeled data or special equipment. In tests, CMIL performed as well as more advanced methods on large datasets and even better than some others. This technology could be useful for identifying people in security cameras or social media photos. |
Keywords
* Artificial intelligence * Supervised
