Summary of On Unsupervised Image-to-image Translation and Gan Stability, by Bahaaeddin Alaila et al.
On Unsupervised Image-to-image translation and GAN stability
by BahaaEddin AlAila, Zahra Jandaghi, Abolfazl Farahani, Mohammad Ziad Al-Saad
First submitted to arxiv on: 18 Oct 2023
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Image and Video Processing (eess.IV)
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 research investigates image-to-image translation, a challenging task with significant implications for various computer vision applications like colorization, inpainting, and segmentation. The study focuses on unpaired (unsupervised) translations, which require sophisticated pattern extraction and application. This problem has gained attention in recent years due to successful applications of deep generative models, particularly Generative Adversarial Networks (GANs). In this work, we analyze failure cases of CycleGAN, a seminal paper in the field, and hypothesize that they are related to GAN stability issues. We propose two general models to alleviate these problems and reach the same conclusion about the task being ill-posed, as discussed in recent literature. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Image-to-image translation is a fascinating problem that can greatly impact various computer vision tasks. Researchers have been trying to develop deep generative models, like GANs, to achieve this goal. However, there are some challenges that need to be addressed. One famous paper called CycleGAN had some issues, and scientists have been trying to figure out why. In this research, we look at what went wrong with CycleGAN and suggest ways to make it better. |
Keywords
* Artificial intelligence * Attention * Gan * Translation * Unsupervised
