justin4ai/FD-DINOv2

Unofficial Implementation of Fréchet Distance with DINOv2 backbone in Pytorch.

22
/ 100
Experimental

This tool helps researchers and practitioners evaluate the quality of AI-generated images, particularly from generative models like GANs and diffusion models. It takes a folder of your generated images and a folder of real reference images, then outputs a score indicating how similar and realistic the generated set is compared to the real one. It's designed for anyone working on developing or comparing image generation AI.

No commits in the last 6 months.

Use this if you need a reliable and unbiased metric to assess how realistic and diverse your AI-generated images are, especially when comparing different generative models.

Not ideal if you are looking for an image quality metric that relies on the older Inception-V3 network, or if you need to evaluate individual image quality rather than dataset similarity.

AI image generation Generative models Model evaluation Computer vision research Diffusion models
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 6 / 25
Maturity 16 / 25
Community 0 / 25

How are scores calculated?

Stars

21

Forks

Language

Python

License

Apache-2.0

Last pushed

Jun 30, 2024

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/justin4ai/FD-DINOv2"

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