yunjey/domain-transfer-network
TensorFlow Implementation of Unsupervised Cross-Domain Image Generation
This project helps machine learning engineers and researchers in computer vision. It takes images from one domain (like photos of house numbers) and transforms them to look like images from another domain (like handwritten digits from the MNIST dataset), while keeping the core content the same. This allows you to generate new training data in a target style or format, even if you only have labeled data in a different, source domain.
862 stars. No commits in the last 6 months.
Use this if you need to generate synthetic images that bridge the stylistic gap between different visual datasets, for example, to augment a dataset with diverse appearances.
Not ideal if you need to translate images where the underlying content or labels should also change, rather than just the visual style.
Stars
862
Forks
200
Language
Python
License
MIT
Category
Last pushed
Jun 06, 2018
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/yunjey/domain-transfer-network"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Related models
taesungp/contrastive-unpaired-translation
Contrastive unpaired image-to-image translation, faster and lighter training than cyclegan (ECCV...
PaddlePaddle/PaddleGAN
PaddlePaddle GAN library, including lots of interesting applications like First-Order motion...
tohinz/ConSinGAN
PyTorch implementation of "Improved Techniques for Training Single-Image GANs" (WACV-21)
sagiebenaim/DistanceGAN
Pytorch implementation of "One-Sided Unsupervised Domain Mapping" NIPS 2017
weihaox/awesome-image-translation
Collection of awesome resources on image-to-image translation.