https://chenshuang-zhang.github.io/imagenet_d/

[CVPR2024 Highlight] Official Code for "ImageNet-D: Benchmarking Neural Network Robustness on Diffusion Synthetic Object"
https://chenshuang-zhang.github.io/imagenet_d/

benchmark computer-vision dataset diffusion-models generative-models image-recognition imagenet large-language-model multi-modality out-of-distribution recognition robustness stable-diffusion synthetic-data text-to-image-synthesis vision-language-model

Last synced: about 2 months ago
JSON representation

[CVPR2024 Highlight] Official Code for "ImageNet-D: Benchmarking Neural Network Robustness on Diffusion Synthetic Object"

• Host: GitHub
• URL: https://chenshuang-zhang.github.io/imagenet_d/
• Owner: chenshuang-zhang
• License: mit
• Created: 2024-03-14T07:39:22.000Z (6 months ago)
• Default Branch: main
• Last Pushed: 2024-05-02T05:44:08.000Z (5 months ago)
• Last Synced: 2024-06-17T00:28:20.247Z (3 months ago)
• Topics: benchmark, computer-vision, dataset, diffusion-models, generative-models, image-recognition, imagenet, large-language-model, multi-modality, out-of-distribution, recognition, robustness, stable-diffusion, synthetic-data, text-to-image-synthesis, vision-language-model
• Language: Python
• Homepage: https://arxiv.org/abs/2403.18775
• Size: 49.3 MB
• Stars: 33
• Watchers: 2
• Forks: 5
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• awesome-cvpr-2024 - ImageNet-D: Benchmarking Neural Network Robustness on Diffusion Synthetic Object - zhang/imagenet_d?style=social)](https://github.com/chenshuang-zhang/imagenet_d) [](https://arxiv.org/html/2403.18775v1) | ImageNet-D is a new benchmark for evaluating neural network robustness in visual perception tasks. It generates synthetic images with diverse backgrounds, textures, and materials, making it more challenging than other synthetic datasets. Key features include diversified image generation, high visual fidelity, and significant accuracy reduction of various vision models. The benchmark is created by combining object categories and refining through human verification. ImageNet-D is effective in evaluating neural network robustness, as accuracy on it improves with accuracy on ImageNet. | (📊 Datasets/Benchmarks)