/Image-Translation-Models

Primary LanguagePython

Collections of Image Translation Models

python pytorch pytorch_lignthing hydra black

Structurally configured framework for typical generative methods with different networks and datasets.



Implemented Models

Supervised Methods

  • ✓ pix2pix

  • ❏ pix2pixHD

  • ❏ SPADE

Unsupervised Methods

  • ❏ CycleGAN

  • ❏ UNIT

  • ❏ MUNIT

Getting Started

Installation

  • Clone this repo

git clone https://github.com/Victarry/Image-Translation-Models.git
cd Image-Translation-Models

  • Install Pytorch >1.8.0 and other dependent packages.

    • For pip users, use pip install -r requirements.txt

    • For conda users, use conda env create -f environment.yml

Data Preparation

Download pix2pix dataset from http://efrosgans.eecs.berkeley.edu/pix2pix/datasets/ and extract into datasets folder.

Model Traning

  • Train a model

python run.py model=pix2pix datamodule=aligned dataset=night2day networks=res_conv

  • You can specify the logging dir by setting exp_name like exp_name=pix2pix/exp1 in above command line.

  • Use tensorboard to visualize experiment results in logs/runs

tensorboard --logdir logs/runs