by Binhui Xie, Shuang Li, Fangrui Lv, Chi Harold Liu, Guoren Wang, and Dapeng Wu
This repo contains the official PyTorch code and models for the CAF.
Update on 2022/06/24: Paper is available under the "Early Access" area on IEEE Xplore.
In this work, we propose a unified framework, called Collaborative Alignment Framework (CAF), which simultaneously reduces the global domain discrepancy and preserves the local semantic consistency for cross-domain knowledge transfer in a collaborative manner.
If you find this project useful in your research, please consider citing:
@ARTICLE{xie2022caf,
author={Binhui Xie, Shuang Li, Fangrui Lv, Chi Harold Liu, Guoren Wang, and Dapeng Wu},
journal={IEEE Transactions on Knowledge and Data Engineering},
title={A Collaborative Alignment Framework of Transferable Knowledge Extraction for Unsupervised Domain Adaptation},
year={2022},
volume={},
number={},
pages={1-1},
doi={10.1109/TKDE.2021.3118111}
}
For this project, we used python 3.7.5. We recommend setting up a new virtual environment:
Step-by-step installation
conda create --name CAF -y python=3.7
conda activate CAF
# this installs the right pip and dependencies for the fresh python
conda install -y ipython pip
pip install -r requirements.txt
- Download The DomainNet Dataset (cleaned version)
- Download The VisDA-2017 Dataset
- Download The Office-31 Dataset
- Download The ImageCLEF Dataset
The data folder should be structured as follows:
├── data/
│ ├── domainnet/
| | ├── clipart/
| | ├── infograph/
| | ├── painting/
| | ├── quickdraw/
| | ├── real/
| | ├── sketch/
│ ├── visda2017/clf/
| | ├── train/
| | ├── validation/
│ ├── office31/
| | ├── amazon/
| | ├── dslr/
| | ├── webcam/
│ ├── imageCLEF/
| | ├── c/
| | ├── i/
| | ├── p/
│ └──
Symlink the required dataset
ln -s /path_to_domainnet_dataset/ data/domainnet
ln -s /path_to_visda2017_dataset/clf/ data/visda2017
ln -s /path_to_office31_dataset data/office31
ln -s /path_to_imageCLEF_dataset data/imageCLEF
For DomainNet (Table 2)
bash scripts/domainnet_res50.sh
bash scripts/domainnet_res101.sh
For VisDA2017 (Table 3)
bash scripts/visda.sh
For Office31 (Table 4)
bash scripts/office.sh
For ImageCLEF (Table 5)
bash scripts/celf.sh
This project is based on the following open-source projects. We thank their authors for making the source code publicly available.
If you have any problem about our code, feel free to contact
or describe your problem in Issues.