Awesome

CroDoBo

Introduction

• Pytorch Implementation of Burn After Reading: Online Adaptation for Cross-domain Streaming Data

Environment

• Code is developed with cuda 11.4, python 3.8.5

Requirements

• torch 1.8.1
• wilds 1.1.0
• pillow

Dataset Preparation

• VisDA-C </br>
• COVID-DA</br>
• WILDS</br>
• Fashion-MNIST</br>
• DeepFashion</br> Prepare the randomly perturbated data by python gen_random_perm.py

Training

python main_crodobo.py --backbone FEATURENETWORK --dataset DATASETOPTION --data_root PATH/TO/YOUR/DATASET

Contact

• Please email loyo@umd.edu or cramaiah@salesforce.com if you have any questions.

Citation

If you find this codebase useful, please cite our paper:

@article{yang2021burn,
  title={Burn After Reading: Online Adaptation for Cross-domain Streaming Data},
  author={Yang, Luyu and Gao, Mingfei and Chen, Zeyuan and Xu, Ran and Shrivastava, Abhinav and Ramaiah, Chetan},
  journal={arXiv preprint arXiv:2112.04345},
  year={2021}
}

Acknowledgement

• We referred to SHOT for the implementation.

License

Our code is BSD-3 licensed. See LICENSE.txt for details.