chenshuang-zhang/imagenet_d

[CVPR 2024 Highlight] ImageNet-D

34
/ 100
Emerging

ImageNet-D provides a specialized dataset to evaluate how well computer vision models, from classic classifiers like ResNet to advanced foundation models like CLIP and MiniGPT-4, can recognize objects in images generated by AI diffusion models. It takes a trained vision model and synthetic images as input, and outputs its object recognition accuracy under various challenging conditions like altered backgrounds, textures, or materials. This is for researchers and engineers developing and testing robust computer vision systems.

No commits in the last 6 months.

Use this if you need to rigorously test the real-world robustness of your image classification, vision-language, or visual question answering models against highly varied and realistic synthetic data.

Not ideal if you are looking for a standard image classification dataset or if your primary focus is on training models from scratch rather than benchmarking their robustness.

computer-vision-benchmarking model-robustness-testing synthetic-data-evaluation deep-learning-research AI-model-auditing
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 8 / 25
Maturity 16 / 25
Community 10 / 25

How are scores calculated?

Stars

47

Forks

5

Language

Python

License

MIT

Last pushed

Oct 15, 2024

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/chenshuang-zhang/imagenet_d"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.