Self-Challenging Improves Cross-Domain Generalization (In progress)

This is the official implementation of:

Zeyi Huang, Haohan Wang, Eric P. Xing, and Dong Huang, Self-Challenging Improves Cross-Domain Generalization, ECCV, 2020 (Oral), arxiv version.

Citation:

@inproceedings{huangRSC2020,
  title={Self-Challenging Improves Cross-Domain Generalization},
  author={Zeyi Huang and Haohan Wang and Eric P. Xing and Dong Huang},
  booktitle={ECCV},
  year={2020}
}

Installation

Requirements:

Python ==3.7
Pytorch >=1.0.0
Torchvision >= 0.2.0
Cuda >=10.0
Tensorflow >=1.14
GPU: RTX 2080

Data Preparation

Download PACS dataset from here. Once you have download the data, you must update the files in data/txt_list to match the actual location of your files.

Step-by-step installation

Runing on PACS dataset

Experiments with different source/target domains are listed in train.py(L140-147). To train a ResNet18, simply run:

  python train.py

Other pretrained models

New ImageNet ResNet baselines training by RSC.

Backbone	Top-1 Acc %	Top-5 Acc %	pth models
ResNet-50	77.18	93.53	download
ResNet-101	78.23	94.16	download
ResNet-152	78.89	94.43	download

Acknowledgement

We borrowed some code and data augmentation techniques from Jigen.

youtang1993/RSC