snap-research/CAT
[CVPR 2021] Teachers Do More Than Teach: Compressing Image-to-Image Models (CAT)
This project helps creators and developers reduce the size of image-to-image AI models used for tasks like changing an image from day to night, applying artistic styles, or converting sketches to realistic photos. It takes a large, pre-trained image transformation model and outputs a significantly smaller, more efficient version suitable for mobile devices or applications where performance is critical. This is ideal for those building interactive visual applications or augmented reality experiences on platforms like Snapchat.
182 stars. No commits in the last 6 months.
Use this if you need to deploy complex image style transfer or transformation AI models directly onto mobile phones or other resource-constrained environments, ensuring they run smoothly without significant latency.
Not ideal if you are developing a new image transformation model from scratch or if your application runs on powerful servers with ample computational resources.
Stars
182
Forks
20
Language
Python
License
—
Last pushed
Apr 22, 2022
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/snap-research/CAT"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.