This Repo is still under Construction!! There may be issues!

First, Create an Environment

conda create -n domain_ada 
conda activate domain_ada

pip3 install torch==1.8.1+cu111 torchvision==0.9.1+cu111 -f https://download.pytorch.org/whl/torch_stable.html
pip install hydra-core numpy omegaconf sklearn tqdm wandb

Download the dataset: VISDA-C

1. For VISDA-C, go to this link Link and download the train.tar, validation.tar and test.tar

Put these downloaded files into "data/VISDA-C/" folder. The text files should be in the proper folder, expercially validation_list.txt file.

Download the source model

1. For Downloading the source models trained on VISDA-C from here Link

2. Put them in the "checkpoint" folder.

Run Commands

1. For VISDA-C dataset, for adapting a model from "train" to "val"

    export CUDA_VISIBLE_DEVICES=0,1,2,3
    bash train_target_VisDA.sh 
   

To check on our proposed algorithm, please go to "target_visda.py" and check "train_csfda" function.

Note: 1. If the simulation ends without any error, set HYDRA_FULL_ERROR=1 GPU Usage: 2 NVIDIA RTX A40 GPUs