Summary of Imagenet-d: Benchmarking Neural Network Robustness on Diffusion Synthetic Object, by Chenshuang Zhang et al.
ImageNet-D: Benchmarking Neural Network Robustness on Diffusion Synthetic Object
by Chenshuang Zhang, Fei Pan, Junmo Kim, In So Kweon, Chengzhi Mao
First submitted to arxiv on: 27 Mar 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Artificial Intelligence (cs.AI); 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 paper introduces a new benchmark for visual perception robustness, called ImageNet-D, which generates high-quality synthetic images with diversified backgrounds, textures, and materials. This benchmark is designed to challenge the accuracy of deep learning models, including popular vision models like ResNet, CLIP, and MiniGPT-4, by reducing their performance by up to 60%. The paper leverages diffusion models to generate these challenging images and shows that they are effective in testing vision models. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The researchers create a new benchmark for visual perception robustness called ImageNet-D. They use a special type of computer model called a “diffusion model” to make the pictures look more real. This helps test how well computers can understand what’s in pictures, even if they’re tricky or noisy. The results show that popular computer models are not very good at understanding these challenging images. |
Keywords
* Artificial intelligence * Deep learning * Diffusion model * Resnet
