Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/mit-han-lab/data-efficient-gans

[NeurIPS 2020] Differentiable Augmentation for Data-Efficient GAN Training
https://github.com/mit-han-lab/data-efficient-gans

data-efficient gans generative-adversarial-network image-generation neurips-2020 pytorch tensorflow

Last synced: 29 days ago
JSON representation

[NeurIPS 2020] Differentiable Augmentation for Data-Efficient GAN Training

Host: GitHub
URL: https://github.com/mit-han-lab/data-efficient-gans
Owner: mit-han-lab
License: bsd-2-clause
Created: 2020-06-15T16:07:24.000Z (over 4 years ago)
Default Branch: master
Last Pushed: 2024-09-24T17:46:07.000Z (about 2 months ago)
Last Synced: 2024-10-01T20:40:29.063Z (about 1 month ago)
Topics: data-efficient, gans, generative-adversarial-network, image-generation, neurips-2020, pytorch, tensorflow
Language: Python
Homepage: https://arxiv.org/abs/2006.10738
Size: 33.9 MB
Stars: 1,274
Watchers: 19
Forks: 175
Open Issues: 27
Metadata Files:
- Readme: README.md
- License: LICENSE.txt

Awesome Lists containing this project

README

        # Data-Efficient GANs with DiffAugment

### [project](https://data-efficient-gans.mit.edu/) | [paper](https://arxiv.org/pdf/2006.10738) | [datasets](https://www.dropbox.com/scl/fi/f641djlqsyen5emxnh22r/data-efficient-gans.zip?rlkey=xlmkf445fzwfno91i7954sp60&st=0itvqzdc&dl=0) | [video](https://www.youtube.com/watch?v=SsqcjS6SVM0) | [slides](https://data-efficient-gans.mit.edu/slides.pdf)



*Generated using only 100 images of Obama, grumpy cats, pandas, the Bridge of Sighs, the Medici Fountain, the Temple of Heaven, without pre-training.*

**[NEW!]** PyTorch training with [DiffAugment-stylegan2-pytorch](https://github.com/mit-han-lab/data-efficient-gans/tree/master/DiffAugment-stylegan2-pytorch) is now available!

**[NEW!]** Our [Colab tutorial](https://colab.research.google.com/gist/zsyzzsoft/5fbb71b9bf9a3217576bebae5de46fc2/data-efficient-gans.ipynb) is released! [![](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/gist/zsyzzsoft/5fbb71b9bf9a3217576bebae5de46fc2/data-efficient-gans.ipynb)

**[NEW!]** FFHQ training is supported! See the [DiffAugment-stylegan2](https://github.com/mit-han-lab/data-efficient-gans/tree/master/DiffAugment-stylegan2#FFHQ) README.

**[NEW!]** Time to generate 100-shot interpolation videos with [generate_gif.py](https://github.com/mit-han-lab/data-efficient-gans/tree/master/DiffAugment-stylegan2/generate_gif.py)!

**[NEW!]** Our [DiffAugment-biggan-imagenet](https://github.com/mit-han-lab/data-efficient-gans/tree/master/DiffAugment-biggan-imagenet) repo (for TPU training) is released!

**[NEW!]** Our [DiffAugment-biggan-cifar](https://github.com/mit-han-lab/data-efficient-gans/tree/master/DiffAugment-biggan-cifar) PyTorch repo is released!

This repository contains our implementation of Differentiable Augmentation (DiffAugment) in both PyTorch and TensorFlow. It can be used to significantly improve the data efficiency for GAN training. We have provided [DiffAugment-stylegan2](https://github.com/mit-han-lab/data-efficient-gans/tree/master/DiffAugment-stylegan2) (TensorFlow) and [DiffAugment-stylegan2-pytorch](https://github.com/mit-han-lab/data-efficient-gans/tree/master/DiffAugment-stylegan2-pytorch), [DiffAugment-biggan-cifar](https://github.com/mit-han-lab/data-efficient-gans/tree/master/DiffAugment-biggan-cifar) (PyTorch) for GPU training, and [DiffAugment-biggan-imagenet](https://github.com/mit-han-lab/data-efficient-gans/tree/master/DiffAugment-biggan-imagenet) (TensorFlow) for TPU training.



*Low-shot generation without pre-training. With DiffAugment, our model can generate high-fidelity images using only 100 Obama portraits, grumpy cats, or pandas from our collected 100-shot datasets, 160 cats or 389 dogs from the AnimalFace dataset at 256×256 resolution.*



*Unconditional generation results on CIFAR-10. StyleGAN2’s performance drastically degrades given less training data. With DiffAugment, we are able to roughly match its FID and outperform its Inception Score (IS) using only **20%** training data.*

Differentiable Augmentation for Data-Efficient GAN Training


[Shengyu Zhao](https://scholar.google.com/citations?user=gLCdw70AAAAJ), [Zhijian Liu](http://zhijianliu.com/), [Ji Lin](http://linji.me/), [Jun-Yan Zhu](https://www.cs.cmu.edu/~junyanz/), and [Song Han](https://songhan.mit.edu/)


MIT, Tsinghua University, Adobe Research, CMU


[arXiv](https://arxiv.org/pdf/2006.10738.pdf)

## Overview



*Overview of DiffAugment for updating D (left) and G (right). DiffAugment applies the augmentation T to both the real sample x and the generated output G(z). When we update G, gradients need to be back-propagated through T (iii), which requires T to be differentiable w.r.t. the input.*

## Training and Generation with 100 Images [![](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/gist/zsyzzsoft/5fbb71b9bf9a3217576bebae5de46fc2/data-efficient-gans.ipynb)

To generate an interpolation video using our pre-trained models:

```bash

cd DiffAugment-stylegan2

python generate_gif.py -r mit-han-lab:DiffAugment-stylegan2-100-shot-obama.pkl -o obama.gif

```

or to train a new model:

```bash

python run_low_shot.py --dataset=100-shot-obama --num-gpus=4

```

You may also try out `100-shot-grumpy_cat`, `100-shot-panda`, `100-shot-bridge_of_sighs`, `100-shot-medici_fountain`, `100-shot-temple_of_heaven`, `100-shot-wuzhen`, or the folder containing your own training images. Please refer to the [DiffAugment-stylegan2](https://github.com/mit-han-lab/data-efficient-gans/tree/master/DiffAugment-stylegan2#100-shot-generation) README for the dependencies and details.

**[NEW!]** PyTorch training is now available:

```bash

cd DiffAugment-stylegan2-pytorch

python train.py --outdir=training-runs --data=https://data-efficient-gans.mit.edu/datasets/100-shot-obama.zip --gpus=1

```

## DiffAugment for StyleGAN2

To run *StyleGAN2 + DiffAugment* for unconditional generation on the 100-shot datasets, CIFAR, FFHQ, or LSUN, please refer to the [DiffAugment-stylegan2](https://github.com/mit-han-lab/data-efficient-gans/tree/master/DiffAugment-stylegan2) README or [DiffAugment-stylegan2-pytorch](https://github.com/mit-han-lab/data-efficient-gans/tree/master/DiffAugment-stylegan2-pytorch) for the PyTorch version.

## DiffAugment for BigGAN

Please refer to the [DiffAugment-biggan-cifar](https://github.com/mit-han-lab/data-efficient-gans/tree/master/DiffAugment-biggan-cifar) README to run *BigGAN + DiffAugment* for conditional generation on CIFAR (using GPUs), and the [DiffAugment-biggan-imagenet](https://github.com/mit-han-lab/data-efficient-gans/tree/master/DiffAugment-biggan-imagenet) README to run on ImageNet (using TPUs).

## Using DiffAugment for Your Own Training

To help you use DiffAugment in your own codebase, we provide portable DiffAugment operations of both TensorFlow and PyTorch versions in [DiffAugment_tf.py](https://github.com/mit-han-lab/data-efficient-gans/blob/master/DiffAugment_tf.py) and [DiffAugment_pytorch.py](https://github.com/mit-han-lab/data-efficient-gans/blob/master/DiffAugment_pytorch.py). Generally, DiffAugment can be easily adopted in any model by substituting every *D(x)* with *D(T(x))*, where *x* can be real images or fake images, *D* is the discriminator, and *T* is the DiffAugment operation. For example,

```python

from DiffAugment_pytorch import DiffAugment

# from DiffAugment_tf import DiffAugment

policy = 'color,translation,cutout' # If your dataset is as small as ours (e.g.,

# hundreds of images), we recommend using the strongest Color + Translation + Cutout.

# For large datasets, try using a subset of transformations in ['color', 'translation', 'cutout'].

# Welcome to discover more DiffAugment transformations!

...

# Training loop: update D

reals = sample_real_images() # a batch of real images

z = sample_latent_vectors()

fakes = Generator(z) # a batch of fake images

real_scores = Discriminator(DiffAugment(reals, policy=policy))

fake_scores = Discriminator(DiffAugment(fakes, policy=policy))

# Calculating D's loss based on real_scores and fake_scores...

...

...

# Training loop: update G

z = sample_latent_vectors()

fakes = Generator(z) # a batch of fake images

fake_scores = Discriminator(DiffAugment(fakes, policy=policy))

# Calculating G's loss based on fake_scores...

...

```

We have implemented Color, Translation, and Cutout DiffAugment as visualized below:



## Citation

If you find this code helpful, please cite our paper:

```

@inproceedings{zhao2020diffaugment,

  title={Differentiable Augmentation for Data-Efficient GAN Training},

  author={Zhao, Shengyu and Liu, Zhijian and Lin, Ji and Zhu, Jun-Yan and Han, Song},

  booktitle={Conference on Neural Information Processing Systems (NeurIPS)},

  year={2020}

}

```

## Acknowledgements

We thank NSF Career Award #1943349, MIT-IBM Watson AI Lab, Google, Adobe, and Sony for supporting this research.  Research supported with Cloud TPUs from Google's TensorFlow Research Cloud (TFRC). We thank William S. Peebles and Yijun Li for helpful comments.