Prototypical Partial Optimal Transport for Universal Domain Adaptation

Code for paper "Yucheng Yang, Xiang Gu, Jian Sun, Prototypical Partial Optimal Transport for Universal Domain Adaptation, the 37th AAAI Conference on Artificial Intelligence, 2023".

Prerequisites:

python==3.9
pytorch ==1.12.1
torchvision ==0.13.1
numpy==1.23.1
POT==0.8.2

Datasets:

Download the datasets of
DomainNet
Office-Home
Office-31
VisDA-2017
and modify the path of images in each '.txt' under the folder './data/'.

Training

Office-31:

python train.py --task office31 -s amazon -t dslr --lr 0.0002 --balanced --no-ssl

Office-Home:

python train.py --task officehome -s Art -t Clipart --lr 0.001 --balanced --mlp --aug-plus --cos --multiprocessing-distributed

VisDA-2017:

python train.py --task VisDA2017 -s train -t validation --lr 0.0005 --balanced --mlp --aug-plus --cos --multiprocessing-distributed

Reference code:

https://github.com/facebookresearch/moco

Contact：

If you have any problem, feel free to contect ycyang@stu.xjtu.edu.cn.