Summer to Winter Translation
Implement the change of seasons in the photo, the change of winter and summer landscapes.
- Learning PyTorch
- Review articles and papers on the topic Image-to-Image Translation
- Implement CycleGAN architectures
- Train CycleGAN
The CycleGAN is a technique that involves the automatic training of image-to-image translation models without paired examples. The models are trained in an unsupervised manner using a collection of images from the source and target domain that do not need to be related in any way. Read more: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks CycleGAN GitHub Page
CycleGAN Architecture
Generator Architecture
Discriminator Architecture
Implementation of cycle-consistency loss
Summer2Winter Yosemite dataset consists of 1540 Summer Photos & 1200 Winter Photos with each split into train and test subsets.
Summer2Winter Yosemite dataset
Summer to Winter
Winter to Summer
A set of random images of Kazan
Summer to Winter
Winter to Summer
- Overview of CycleGAN architecture and training
- Image-to-Image Translation using CycleGAN Model
- CycleGAN Summer->Winter Image Translation PyTorch
- Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks
- CS231n Convolutional Neural Networks for Visual Recognition
- Deep Residual Learning for Image Recognition
- Residual blocks — Building blocks of ResNet
- Least Squares Generative Adversarial Networks
- Guide to Pytorch Learning Rate Scheduling
- CycleGAN with Better Cycles