Source

• ROCm compatible version of this repository - CycleGAN-Tensorflow-2

---

---

CycleGAN - Tensorflow 2

Tensorflow 2 implementation of CycleGAN.

Paper: Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks

Author: Jun-Yan Zhu et al.

Exemplar results

summer2winter

row 1: summer -> winter -> reconstructed summer, row 2: winter -> summer -> reconstructed winter

horse2zebra

row 1: horse -> zebra -> reconstructed horse, row 2: zebra -> horse -> reconstructed zebra

apple2orange

row 1: apple -> orange -> reconstructed apple, row 2: orange -> apple -> reconstructed orange

Usage

• Prerequisites

  • Tensorflow 2.0pip install tensorflow-rocm==2.0.0b3
  • (if you meet "tf.summary.histogram fails with TypeError" pip install --upgrade tb-nightly)
  • scikit-image, oyaml, tqdm
  • Python 3.6
  • ROCm 2.7

• Dataset

  • download the summer2winter dataset

    sh ./download_dataset.sh summer2winter_yosemite

  • download the horse2zebra dataset

    sh ./download_dataset.sh horse2zebra

  • see download_dataset.sh for more datasets

• Example of training

  HIP_VISIBLE_DEVICES=0 python3 train.py --dataset summer2winter_yosemite

  • tensorboard for loss visualization

    tensorboard --logdir ./output/summer2winter_yosemite/summaries --port 6006

• Example of testing

  HIP_VISIBLE_DEVICES=0 python3 test.py --experiment_dir ./output/summer2winter_yosemite