FUNIT-pytorch: Few-Shot Unsupervised Image-to-Image Translation (ICCV 2019)

Edited from FUNIT-pytorch

Edition

1. Change Input/Output Logic

Paper | Official PyTorch code

Tensorflow Implementation

Tensorflow Implementation

Usage

├── dataset
   └── YOUR_DATASET_NAME
       ├── train
           ├── class1 (class folder)
               ├── xxx.jpg (class1 image)
               ├── yyy.png
               ├── ...
           ├── class2
               ├── aaa.jpg (class2 image)
               ├── bbb.png
               ├── ...
           ├── class3
           ├── ...
       ├── test
           ├── class1
               ├── zzz.jpg (class1 image)
               ├── www.png
               ├── ...
           ├── class2
               ├── ccc.jpg (class2 image)
               ├── ddd.jpg
               ├── ...

Train

> python main.py --name FUNIT --dataset YOUR_DATASET_NAME --phase train --K 1

Test

> python main.py --name FUNIT --dataset YOUR_DATASET_NAME --phase test --K 1

Architecture

Our result

Paper result