Summary of Anti-forgetting Adaptation For Unsupervised Person Re-identification, by Hao Chen et al.
Anti-Forgetting Adaptation for Unsupervised Person Re-identification
by Hao Chen, Francois Bremond, Nicu Sebe, Shiliang Zhang
First submitted to arxiv on: 22 Nov 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 The proposed Dual-level Joint Adaptation and Anti-forgetting (DJAA) framework aims to incrementally adapt a person re-identification (ReID) model to new domains without forgetting previously acquired knowledge. The method utilizes prototype and instance-level consistency to mitigate forgetting during adaptation, storing representative image samples and corresponding cluster prototypes in a memory buffer updated at each step. By regularizing image-to-image similarity and image-to-prototype similarity with the buffered images and prototypes, the model rehearses old knowledge and improves its generalization ability. Extensive experiments demonstrate the significant improvement of DJAA in anti-forgetting, generalization, and backward-compatible abilities. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper proposes a new method for person re-identification that helps a model learn from one place and remember what it learned from another place. This is important because existing methods are not good at remembering what they learned. The proposed method uses two levels: prototype and instance. It stores some images in memory, updates them as it adapts to new places, and then uses these stored images to make sure the model doesn’t forget what it learned. The results show that this method is much better than existing methods. |
Keywords
* Artificial intelligence * Generalization
