This is the demo for IMVDA via CEKT (updating...).
Office-31 and Office-Home are popular datasets for domain adaptation, which only includes RGB images.
RGB-D and B3DO are public available datasets with RGB and Depth images. Note that the original depth images should be processed and then are fed into the network.
python==3.6.2
pytorch==1.0.0
torchvision==0.2.2
numpy==1.16.2
scikit-learn==0.20.2
scipy==1.1.0
opencv-python==4.0.0
The implementation of channel exchange is mainly based on CEN. Many thanks to Yikai for his previous work.