DAN

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.

    Requirement

  • python 3

  • pytorch 1.0
  • torchvision 0.2.0

    Usage

  1. You can download Office31 dataset here. And then unrar dataset in ./dataset/.
  2. You can change the source_name and target_name in MRAN.py to set different transfer tasks.
  3. Run python MRAN.py.

    Results on Office31

    | Method | A - W | D - W | W - D | A - D | D - A | W - A | Average | | —- | —- | —- | —- | —- | —- | —- | —- |

| MRAN | 91.4±0.1 | 96.9±0.3 | 99.8±0.2 | 86.4±0.6 | 68.3±0.5 | 70.9±0.6 | 85.6 |