This repository contains the code for 'Domain-Agnostic Mutual Prompting for Unsupervised Domain Adaptation' (Accepted to CVPR 2024)
Our code is built based on CLIP and Dassl, which can be installed with following commands.
# install CLIP
pip install git+https://github.com/openai/CLIP.git
# install Dassl
git clone https://github.com/KaiyangZhou/Dassl.pytorch.git
cd dassl
pip install -r requirements.txt
pip install .
cd..
One can install other dependent tools via
pip install -r requirements.txt
The datasets used for UDA tasks can be downloaded via the following links.
VisDA17 (http://ai.bu.edu/visda-2017/#download)
Office-Home (https://drive.google.com/file/d/0B81rNlvomiwed0V1YUxQdC1uOTg/view?resourcekey=0-2SNWq0CDAuWOBRRBL7ZZsw)
Mini-DomainNet (http://ai.bu.edu/DomainNet/)
After downloading the datasets, please update the dataset paths in scripts/{dataset}.sh accordingly.