ml-lab / sg-gan

Semantic-aware Grad-GAN for Virtual-to-Real Urban Scene Adaption

SG-GAN

TensorFlow implementation of SG-GAN.

Prerequisites

• TensorFlow (implemented in v1.3)
• numpy
• scipy
• pillow

Getting Started

Train

• Prepare dataset. We present example images in datasets folder for reference as data format, and scripts prepare_data.py and segment_class.py for reference in preparing dataset.

• Train a model:

CUDA_VISIBLE_DEVICES=0 python main.py

Models are saved to ./checkpoints/ (can be changed by passing --checkpoint_dir=your_dir).

• Continue training a model (useful for updating parameters)

CUDA_VISIBLE_DEVICES=0 python main.py --continue_train 1

Test

• Finally, test the model:

CUDA_VISIBLE_DEVICES=0 python main.py --phase test --img_width 2048 --img_height 1024

Adapted test images will be outputted to ./test/ (can be changed by passing --test_dir=your_dir).

Reference

• The TensorFlow implementation of CycleGAN, https://github.com/xhujoy/CycleGAN-tensorflow