Summary of Learning Image Demoireing From Unpaired Real Data, by Yunshan Zhong et al.
Learning Image Demoireing from Unpaired Real Data
by Yunshan Zhong, Yuyao Zhou, Yuxin Zhang, Fei Chao, Rongrong Ji
First submitted to arxiv on: 5 Jan 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 proposed UnDeM method tackles the issue of image demoireing by learning from unpaired real data, rather than relying on paired real data. This approach synthesizes pseudo moire images from unpaired datasets to generate pairs with clean images for training demoireing models. To achieve this, the authors divide real moire images into patches and group them according to their moire complexity. A novel moire generation framework is introduced to synthesize moire images with diverse moire features and details akin to real moire-free images. Additionally, an adaptive denoise method eliminates low-quality pseudo moire images that can impact the learning of demoireing models. The UnDeM method outperforms existing methods using MBCNN and ESDNet-L demoireing models on the FHDMi and UHDM datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Imagine you’re trying to remove annoying patterns from pictures. Most people try to learn from pairs of clean and messy images, but what if you only have messy images? This paper shows how to use those messy images to train a special kind of computer program that can fix the messiness. The authors break down the messy images into smaller pieces, group them by how complicated they are, and then create new, pretend messy images that look like real ones. They also come up with a way to remove unwanted noise from these pretend images. When they tested their method on some common datasets, it worked better than other methods using special computer programs. |
