Summary of Relation-aware Meta-learning For Zero-shot Sketch-based Image Retrieval, by Yang Liu et al.
Relation-Aware Meta-Learning for Zero-shot Sketch-Based Image Retrieval
by Yang Liu, Jiale Du, Xinbo Gao, Jungong Han
First submitted to arxiv on: 28 Nov 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 The paper proposes a novel framework for Zero-Shot Sketch-Based Image Retrieval (ZS-SBIR), which addresses the limitation of traditional SBIR’s inability to retrieve classes absent from the training set. The proposed framework employs a pair-based relation-aware quadruplet loss to bridge feature gaps and enhance inter-class separability. Additionally, it uses a Relation-Aware Meta-Learning Network (RAMLN) to obtain optimal margin values, which improves the generalization ability of the model. The framework is evaluated on the extended Sketchy and TU-Berlin datasets, demonstrating significant improvements over existing state-of-the-art methods in ZS-SBIR. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper solves a problem with image retrieval from sketches. Currently, this can only be done if the sketch is of something that has been seen before. The new framework allows for sketches of things that have never been seen before to still be matched to images. This is achieved by using a special type of loss function and a unique way of storing information about the features of the sketches and images. The results show that this new method works much better than previous ones. |
Keywords
» Artificial intelligence » Generalization » Loss function » Meta learning » Zero shot
