NikosEfth/freedom

Official PyTorch implementation of the WACV 2025 Oral paper "Composed Image Retrieval for Training-FREE DOMain Conversion".

29
/ 100
Experimental

This tool helps creative professionals and researchers easily find images that match a specific object or scene but in a completely different artistic style or contextual setting. You input an image (like a photo of a cat) and a text description of a desired domain (like "watercolor painting" or "a cat in a jungle"), and it outputs relevant images from a large database that fit both criteria. This is ideal for artists, designers, content creators, or anyone needing to visualize concepts across varying visual styles or environments without extensive training data.

No commits in the last 6 months.

Use this if you need to retrieve images from a vast collection that maintain the essence of a query image but transform its visual style or context based on a text description, without needing to train a new model for each specific domain.

Not ideal if you need to modify specific elements within an image, perform object detection, or if your primary goal is to generate entirely new images rather than retrieve existing ones.

creative-asset-discovery image-style-transfer visual-content-search digital-art contextual-image-retrieval
Stale 6m No Package No Dependents
Maintenance 2 / 25
Adoption 8 / 25
Maturity 16 / 25
Community 3 / 25

How are scores calculated?

Stars

47

Forks

1

Language

Python

License

MIT

Last pushed

Aug 31, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/NikosEfth/freedom"

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