1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
|
<img src='imgs/teaser_720.gif' align="right" width=360>
<br><br><br><br>
# pix2pixHD
### [[Project]](https://tcwang0509.github.io/pix2pixHD/) [[Youtube]](https://youtu.be/3AIpPlzM_qs) [[Paper]](https://arxiv.org/pdf/1711.11585.pdf) <br>
Pytorch implementation of our method for high-resolution (e.g. 2048x1024) photorealistic image-to-image translation. It can be used for turning semantic label maps into photo-realistic images or synthesizing portraits from face label maps. <br><br>
[High-Resolution Image Synthesis and Semantic Manipulation with Conditional GANs](https://tcwang0509.github.io/pix2pixHD/)
[Ting-Chun Wang](https://tcwang0509.github.io/)<sup>1</sup>, [Ming-Yu Liu](http://mingyuliu.net/)<sup>1</sup>, [Jun-Yan Zhu](http://people.eecs.berkeley.edu/~junyanz/)<sup>2</sup>, Andrew Tao<sup>1</sup>, [Jan Kautz](http://jankautz.com/)<sup>1</sup>, [Bryan Catanzaro](http://catanzaro.name/)<sup>1</sup>
<sup>1</sup>NVIDIA Corporation, <sup>2</sup>UC Berkeley
In arxiv, 2017.
## Image-to-image translation at 2k/1k resolution
- Our label-to-streetview results
<p align='center'>
<img src='imgs/teaser_label.png' width='440'/>
<img src='imgs/teaser_ours.jpg' width='440'/>
</p>
- Interactive editing results
<p align='center'>
<img src='imgs/teaser_style.gif' width='440'/>
<img src='imgs/teaser_label.gif' width='440'/>
</p>
- Additional streetview results
<p align='center'>
<img src='imgs/cityscapes_1.jpg' width='440'/>
<img src='imgs/cityscapes_2.jpg' width='440'/>
</p>
<p align='center'>
<img src='imgs/cityscapes_3.jpg' width='440'/>
<img src='imgs/cityscapes_4.jpg' width='440'/>
</p>
- Label-to-face and interactive editing results
<p align='center'>
<img src='imgs/face1_1.jpg' width='290'/>
<img src='imgs/face1_2.jpg' width='290'/>
<img src='imgs/face1_3.jpg' width='290'/>
</p>
<p align='center'>
<img src='imgs/face2_1.jpg' width='290'/>
<img src='imgs/face2_2.jpg' width='290'/>
<img src='imgs/face2_3.jpg' width='290'/>
</p>
- Our editing interface
<p align='center'>
<img src='imgs/city_short.gif' width='380'/>
<img src='imgs/face_short.gif' width='490'/>
</p>
## Prerequisites
- Linux or macOS
- Python 2 or 3
- NVIDIA GPU (12G or 24G memory) + CUDA cuDNN
## Getting Started
### Installation
- Install PyTorch and dependencies from http://pytorch.org
- Install python libraries [dominate](https://github.com/Knio/dominate).
```bash
pip install dominate
```
- Clone this repo:
```bash
git clone https://github.com/NVIDIA/pix2pixHD
cd pix2pixHD
```
### Testing
- A few example Cityscapes test images are included in the `datasets` folder.
- Please download the pre-trained Cityscapes model from [here](https://drive.google.com/file/d/1h9SykUnuZul7J3Nbms2QGH1wa85nbN2-/view?usp=sharing) (google drive link), and put it under `./checkpoints/label2city_1024p/`
- Test the model (`bash ./scripts/test_1024p.sh`):
```bash
#!./scripts/test_1024p.sh
python test.py --name label2city_1024p --netG local --ngf 32 --resize_or_crop none
```
The test results will be saved to a html file here: `./results/label2city_1024p/test_latest/index.html`.
More example scripts can be found in the `scripts` directory.
### Dataset
- We use the Cityscapes dataset. To train a model on the full dataset, please download it from the [official website](https://www.cityscapes-dataset.com/) (registration required).
After downloading, please put it under the `datasets` folder in the same way the example images are provided.
### Training
- Train a model at 1024 x 512 resolution (`bash ./scripts/train_512p.sh`):
```bash
#!./scripts/train_512p.sh
python train.py --name label2city_512p
```
- To view training results, please checkout intermediate results in `./checkpoints/label2city_512p/web/index.html`.
If you have tensorflow installed, you can see tensorboard logs in `./checkpoints/label2city_512p/logs` by adding `--tf_log` to the training scripts.
### Multi-GPU training
- Train a model using multiple GPUs (`bash ./scripts/train_512p_multigpu.sh`):
```bash
#!./scripts/train_512p_multigpu.sh
python train.py --name label2city_512p --batchSize 8 --gpu_ids 0,1,2,3,4,5,6,7
```
Note: this is not tested and we trained our model using single GPU only. Please use at your own discretion.
### Training at full resolution
- To train the images at full resolution (2048 x 1024) requires a GPU with 24G memory (`bash ./scripts/train_1024p_24G.sh`).
If only GPUs with 12G memory are available, please use the 12G script (`bash ./scripts/train_1024p_12G.sh`), which will crop the images during training. Performance is not guaranteed using this script.
### Training with your own dataset
- If you want to train with your own dataset, please generate label maps which are one-channel whose pixel values correspond to the object labels (i.e. 0,1,...,N-1, where N is the number of labels). This is because we need to generate one-hot vectors from the label maps. Please also specity `--label_nc N` during both training and testing.
- If your input is not a label map, please just specify `--label_nc 0` which will directly use the RGB colors as input. The folders should then be named `train_A`, `train_B` instead of `train_label`, `train_img`, where the goal is to translate images from A to B.
- If you don't have instance maps or don't want to use them, please specify `--no_instance`.
- The default setting for preprocessing is `scale_width`, which will scale the width of all training images to `opt.loadSize` (1024) while keeping the aspect ratio. If you want a different setting, please change it by using the `--resize_or_crop` option. For example, `scale_width_and_crop` first resizes the image to have width `opt.loadSize` and then does random cropping of size `(opt.fineSize, opt.fineSize)`. `crop` skips the resizing step and only performs random cropping. If you don't want any preprocessing, please specify `none`, which will do nothing other than making sure the image is divisible by 32.
## More Training/Test Details
- Flags: see `options/train_options.py` and `options/base_options.py` for all the training flags; see `options/test_options.py` and `options/base_options.py` for all the test flags.
- Instance map: we take in both label maps and instance maps as input. If you don't want to use instance maps, please specify the flag `--no_instance`.
## Citation
If you find this useful for your research, please use the following.
```
@article{wang2017highres,
title={High-Resolution Image Synthesis and Semantic Manipulation with Conditional GANs},
author={Ting-Chun Wang and Ming-Yu Liu and Jun-Yan Zhu and Andrew Tao and Jan Kautz and Bryan Catanzaro},
journal={arXiv preprint arXiv:1711.11585},
year={2017}
}
```
## Acknowledgments
This code borrows heavily from [pytorch-CycleGAN-and-pix2pix](https://github.com/junyanz/pytorch-CycleGAN-and-pix2pix).
|
