kkanshul/finegan
FineGAN: Unsupervised Hierarchical Disentanglement for Fine-grained Object Generation and Discovery
This project helps researchers and machine learning engineers synthesize realistic images of objects by breaking down the generation process into steps: first the background, then the object's shape, and finally its appearance. You provide a dataset of images, and it outputs new, distinct images with control over these elements. This is ideal for those working on computer vision research or developing AI models for image generation.
281 stars. No commits in the last 6 months.
Use this if you need to generate a diverse set of synthetic images of specific object categories, with the ability to control background, shape, and appearance independently.
Not ideal if you need a user-friendly application for image editing or if you lack expertise in machine learning model training and configuration.
Stars
281
Forks
45
Language
Python
License
BSD-2-Clause
Category
Last pushed
Aug 02, 2021
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/kkanshul/finegan"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
yunjey/domain-transfer-network
TensorFlow Implementation of Unsupervised Cross-Domain Image Generation
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