xrenaa/DisCo

[ICLR2022] Code for "Learning Disentangled Representation by Exploiting Pretrained Generative Models: A Contrastive Learning View"

29
/ 100
Experimental

DisCo helps AI researchers and practitioners analyze the factors that make up generated images. Given a pre-trained generative model (like GANs, VAEs, or Flow models) and a collection of generated images, it can identify and extract distinct, understandable attributes, such as 'pose' or 'smile' in faces, or 'object color' in shapes. This allows researchers to understand how these models create images and to manipulate specific features.

136 stars. No commits in the last 6 months.

Use this if you need to understand the underlying, independent factors of variation within images generated by a deep learning model, without needing to manually label your data.

Not ideal if you are looking to generate new images from scratch or if you primarily work with real-world images that haven't been synthesized by a generative model.

AI research generative models image analysis feature extraction unsupervised learning
No License Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 10 / 25
Maturity 8 / 25
Community 11 / 25

How are scores calculated?

Stars

136

Forks

10

Language

Python

License

Last pushed

Mar 24, 2022

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/xrenaa/DisCo"

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