https://github.com/pfnet-research/chainer-gan-lib
Chainer implementation of recent GAN variants
https://github.com/pfnet-research/chainer-gan-lib
dcgan deep-learning generative-adversarial-network wgan-gp
Last synced: 9 days ago
JSON representation
Chainer implementation of recent GAN variants
- Host: GitHub
- URL: https://github.com/pfnet-research/chainer-gan-lib
- Owner: pfnet-research
- License: mit
- Created: 2017-07-10T08:06:02.000Z (over 8 years ago)
- Default Branch: master
- Last Pushed: 2023-03-24T23:41:30.000Z (over 2 years ago)
- Last Synced: 2025-05-25T12:06:41.449Z (5 months ago)
- Topics: dcgan, deep-learning, generative-adversarial-network, wgan-gp
- Language: Python
- Size: 69.5 MB
- Stars: 408
- Watchers: 31
- Forks: 84
- Open Issues: 14
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Chainer-GAN-lib
This repository collects chainer implementation of state-of-the-art GAN algorithms.
These codes are evaluated with the _inception score_ on Cifar-10 dataset.
Note that our codes are not faithful re-implementation of the original paper.
How to use
-------
Install the requirements first:
```
pip install -r requirements.txt
```
This implementation has been tested with the following versions.
```
python 3.5.2
chainer 4.0.0
+ https://github.com/chainer/chainer/pull/3615
+ https://github.com/chainer/chainer/pull/3581
cupy 3.0.0
tensorflow 1.2.0 # only for downloading inception model
numpy 1.11.1
```
Download the inception score module forked from [https://github.com/hvy/chainer-inception-score](https://github.com/hvy/chainer-inception-score).
```
git submodule update -i
```
Download the inception model.
```
cd common/inception
python download.py --outfile inception_score.model
```
You can start training with `train.py`.
`python train.py --gpu 0 --algorithm dcgan --out result_dcgan`
Please see `example.sh` to train other algorithms.
Quantitative evaluation
-------
| | [Inception](https://arxiv.org/abs/1606.03498) | Inception (Official) | [FID](https://arxiv.org/abs/1706.08500) |
| ------------- | ------------- | ------------- | ------------- |
| Real data | 12.0 | 11.24 | 3.2 (train vs test) |
| [Progressive](http://research.nvidia.com/publication/2017-10_Progressive-Growing-of) | 8.5 | 8.8 | 19.1 |
| [SN-DCGAN](https://drive.google.com/file/d/0B8HZ50DPgR3eSVV6YlF3XzQxSjQ/view?usp=sharing) | 7.5 | 7.41 | 23.6 |
| [WGAN-GP](https://arxiv.org/abs/1704.00028) | 6.8 | 7.86 (ResNet) | 28.2 |
| [DFM](https://openreview.net/pdf?id=S1X7nhsxl) | 7.3 | 7.72 | 30.1 |
| [Cramer GAN](https://arxiv.org/abs/1705.10743) | 6.4 | - | 32.7 |
| [DRAGAN](https://arxiv.org/abs/1705.07215) | 7.1 | 6.90 | 31.5 |
| [DCGAN-vanilla](https://arxiv.org/abs/1511.06434) | 6.7 | 6.16 [WGAN2] 6.99 [DRAGAN] | 34.3 |
| [minibatch discrimination](https://arxiv.org/abs/1606.03498) | 7.0 | 6.86 (-L+HA) | 31.3 |
| [BEGAN](https://arxiv.org/abs/1703.10717) | 5.4 | 5.62 | 84.0 |
Inception scores are calculated by average of 10 evaluation with 5000 samples.
FIDs are calculated with 50000 train dataset and 10000 generated samples.
Generated images
-------
- Progressive
- SN-DCGAN
- WGAN-GP
- DFM
- Cramer GAN
- DRAGAN
- DCGAN
- Minibatch discrimination
- BEGAN
License
-------
MIT License. Please see the LICENSE file for details.