/DisClusterDA

Code release for Unsupervised Domain Adaptation via Distilled Discriminative Clustering published by Pattern Recognition in 2022

Primary LanguagePythonMIT LicenseMIT

DisClusterDA

Code release for Unsupervised Domain Adaptation via Distilled Discriminative Clustering published by Pattern Recognition in 2022.

Project Page $\cdot$ PDF Download

The paper is available here.

Requirements

  • python 3.6.4
  • pytorch 1.4.0
  • torchvision 0.5.0

Data preparation

The structure of the used datasets is shown in the folder ./data/datasets/ here.

The original datasets can be downloaded here.

Model training

  1. Replace paths and domains in run.sh with those in one's own system.
  2. Install necessary python packages.
  3. Run command sh run.sh.

The results are saved in the folder ./checkpoints/.

Article citation

@article{DisClusterDA,
author = {Hui Tang and Yaowei Wang and Kui Jia},
title = {Unsupervised domain adaptation via distilled discriminative clustering},
journal = {Pattern Recognition},
volume = {127},
pages = {108638},
year = {2022},
issn = {0031-3203},
doi = {https://doi.org/10.1016/j.patcog.2022.108638},
url = {https://www.sciencedirect.com/science/article/pii/S0031320322001194},
}