Yushi-Hu/tifa

TIFA: Accurate and Interpretable Text-to-Image Faithfulness Evaluation with Question Answering

37
/ 100
Emerging

This tool helps researchers and developers working with text-to-image models understand how accurately generated images reflect their input text. You provide text prompts and the images your model creates, and it returns a detailed score and explanation of how well the image aligns with the text. This is for AI researchers or machine learning engineers who need to benchmark and improve their image generation models.

182 stars. No commits in the last 6 months.

Use this if you need an objective, detailed, and interpretable way to evaluate whether your text-to-image model is truly creating images that match the given text descriptions.

Not ideal if you are a casual user generating images and simply want to know if an image looks good, rather than conducting a fine-grained evaluation of model faithfulness.

AI-model-evaluation image-generation text-to-image computer-vision generative-AI
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 11 / 25

How are scores calculated?

Stars

182

Forks

12

Language

Python

License

Apache-2.0

Last pushed

Apr 29, 2024

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/Yushi-Hu/tifa"

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