This repository offers a straightforward and well-documented implementation of CycleGAN in PyTorch. CycleGAN is a deep learning model that enables image-to-image translation without paired data, making it ideal for tasks like photo enhancement, style transfer, and domain adaptation.
Key Features
Unpaired Image Translation: Facilitates transformation between two image domains without the need for paired datasets.
Modular Codebase: Organized structure with separate files for models, datasets, utilities, and training scripts.
Visdom Integration: Supports real-time visualization of training progress and generated images via Visdom.
Predefined Datasets: Includes scripts to download and set up popular datasets such as horse2zebra, monet2photo, and summer2winter_yosemite.
Training and Testing Scripts: Provides clear commands for training and testing the model, along with options for GPU acceleration.
Seerah Legacy
.png)
.png)
.png)
.png)
.png)
.png)
