A PyTorch implementation of 'Multi-representationadaptationnetworkforcross-domainimage classification'. The contributions of this paper are summarized as follows.
- We are the first to learn multiple different domain-invariant representations by Inception Adaptation Module (IAM) for cross-domain image classification.
- A novel Multi-Representation Adaptation Network (MRAN) is proposed to align distributions of multiple different representations which might contain more information about the images.
- python 3
- pytorch 0.3.1
- torchvision 0.2.0
- You can download Office31 dataset here. And then unrar dataset in ./dataset/.
- You can change the
MRAN.pyto set different transfer tasks.
Results on Office31
|Method||A - W||D - W||W - D||A - D||D - A||W - A||Average|
Zhu Y, Zhuang F, Wang J, et al. Multi-representation adaptation network for cross-domain image classification[J]. Neural Networks, 2019.
If you have a problem with the code, please contact firstname.lastname@example.org.