Skip to content

Latest commit

 

History

History
31 lines (17 loc) · 674 Bytes

README.md

File metadata and controls

31 lines (17 loc) · 674 Bytes

Semantically Multi-modal Image Synthesis

Unofficical Pytorch implementation of the papaer Semantically Multi-modal Image Synthesis - Zhen Zhu, Zhiliang Xu, Ansheng You, and Xiang Bai in CVPR 2020.

Official impelemntation will be released soon.

Instalation

  • python 3.7
  • pytorch 1.4.0
python3 -m pip install -r requirements.txt

Prepare dataest

Download Cityscapes Dataset from (https://www.cityscapes-dataset.com/)

Train

python3 main_train.py --data_root ~/dataset/cityscapes --batch_size 1

Results

Save output images to directory checkpoint/test/