Yutong-Zhou-cv/Awesome-Text-to-Image

What is this?

Awesome Text-to-Image is a curated list of resources for people working on AI that turns text descriptions into images (for example, typing "a red cat on a sofa" and getting a picture of that). It collects research papers, datasets, evaluation tools, and project links — all in one place.

What's inside?

The repo organizes resources into several sections:

• Papers with code — grouped by year (2022, 2023, 2024, 2025). The 2022 list alone has 69 papers; the survey section has 11 papers.
• Datasets — collections of images and text used to train or test these systems.
• Quantitative evaluation metrics — ways to measure how good a generated image is.
• Projects — links to actual implementations or demos.
• Special curated lists: a "Best Collection" (hand-picked highlights) and a "Recently Focused Papers" list.

Version status

• Version 2.0 is the current version (updated February 29, 2024). The maintainer says "Paper With Code" and "Other Related Works" sections will be updated in March 2024.
• Version 1.0 (an older, all-in-one layout) is still available but no longer updated as of February 29, 2024.

Origin and maintenance

• Created on October 13, 2020 by Yutong Zhou.
• The maintainer also wrote a survey paper titled "Vision + Language Applications: A Survey" (published at CVPRW 2023), and the repo asks users to cite that paper if they find the list helpful.

Who is it for?

Researchers, students, or developers who want a quick, organized reference to the latest work in text-to-image AI — instead of searching through many separate sources.

How to use it

Browse the table of contents, click a section (e.g., "2022" papers), and you'll find a list of papers with links to their code or project pages. It's a reference index, not a tool you run yourself.