GaParmar/clean-fid

PyTorch - FID calculation with proper image resizing and quantization steps [CVPR 2022]

54
/ 100
Established

This tool helps researchers and practitioners evaluate the quality of images generated by AI models. You input a folder of generated images and a folder of real images (or a reference dataset), and it outputs a reliable Fréchet Inception Distance (FID) or Kernel Inception Distance (KID) score. This is for anyone working with generative adversarial networks (GANs) or similar image generation models who needs consistent and accurate evaluation metrics.

1,140 stars. No commits in the last 6 months. Available on PyPI.

Use this if you need to precisely and consistently compare the quality of images produced by different generative AI models, ensuring that variations in evaluation scores are due to model performance, not technical inconsistencies in calculation methods.

Not ideal if you are looking for a tool to generate images or to perform basic image processing tasks like resizing or format conversion.

generative-ai image-synthesis model-evaluation computer-vision deep-learning-research
Stale 6m
Maintenance 2 / 25
Adoption 10 / 25
Maturity 25 / 25
Community 17 / 25

How are scores calculated?

Stars

1,140

Forks

80

Language

Python

License

MIT

Last pushed

Aug 02, 2025

Commits (30d)

0

Dependencies

7

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/GaParmar/clean-fid"

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