https://github.com/zhiqiangdon/online-augment

Code for "OnlineAugment: Online Data Augmentation with Less Domain Knowledge" (ECCV 2020)
https://github.com/zhiqiangdon/online-augment

auto-augment autoaugment automl data-augmentation efficient-training generative-adversarial-network less-domain-knowledge online-augment online-learning pytorch stn vae

Last synced: 14 days ago
JSON representation

Code for "OnlineAugment: Online Data Augmentation with Less Domain Knowledge" (ECCV 2020)

• Host: GitHub
• URL: https://github.com/zhiqiangdon/online-augment
• Owner: zhiqiangdon
• License: apache-2.0
• Created: 2020-07-13T05:18:49.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2020-08-23T03:03:45.000Z (about 4 years ago)
• Last Synced: 2024-10-03T12:36:03.900Z (about 1 month ago)
• Topics: auto-augment, autoaugment, automl, data-augmentation, efficient-training, generative-adversarial-network, less-domain-knowledge, online-augment, online-learning, pytorch, stn, vae
• Language: Python
• Homepage:
• Size: 20 MB
• Stars: 37
• Watchers: 2
• Forks: 4
• Open Issues: 2
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# OnlineAugment (Accepted at ECCV 2020)

Official [OnlineAugment](https://arxiv.org/abs/2007.09271) implementation in PyTorch

- More automatic than AutoAugment and related 

  - Towards fully automatic (STN and VAE, No need to specify the image primitives). 

  - Broad domains (natural, medical images, etc). 

  - Diverse tasks (classification, segmentation, etc). 

- Easy to use 

  - One-stage training (user-friendly). 

  - Simple code (single GPU training, no need for parallel optimization). 

- Orthogonal to AutoAugment and related 

  - Online v.s. Offline (Joint optimization, no expensive offline policy searching). 

  - State-of-the-art performance (in combination with AutoAugment). 

  

 ![](./vis/framework.png)

 

 (In this implementation, we disable the meta-gradient for efficient training. The code is also refactored accordingly, achieving comparable performance. Especially for reduced CIFARs, we observe higher accuracy than reported in the paper.)

## Visualization on CIFAR-10

A-STN

![](./vis/STN.gif)

D-VAE

![](./vis/deform.gif)

P-VAE

![](./vis/texture.gif)

## Run

We conducted experiments in

- python 3.7

- pytorch 1.2, torchvision 0.4.0, cuda10

The searching of policies and training of target model is optimized jointly. 

For example, training wide-resnet using STN on reduced CIFAR-10, using the script in r-cifar10-wrn-scripts

```

./run-aug-stn.sh

```

## Citation

If this code is helpful for your research, please cite:

```

@article{tang2020onlineaugment,

  title={OnlineAugment: Online Data Augmentation with Less Domain Knowledge},

  author={Tang, Zhiqiang and Gao, Yunhe and Karlinsky, Leonid and Sattigeri, Prasanna and Feris, Rogerio and Metaxas, Dimitris},

  journal={arXiv preprint arXiv:2007.09271},

  year={2020}

}

```