Dual Prompt Coreset with Memory for Continual Test-Time Adaptation

Overview

Experiments

Environment

conda create -n CTTA python==3.9.7
conda activate CTTA
pip install torch==2.1.1 torchvision==0.16.1 torchaudio==2.1.1 --index-url https://download.pytorch.org/whl/cu121
pip install -r requirements.txt

Data

• ImageNet-C Download
• Cifar10-to-Cifar10-C Automatic Download
• Cifar100-to-Cifar100-C Automatic Download

Source Domain Statistics

You can set --train_info to use the provided statistics.

• ImageNet-C Download
• Cifar10-to-Cifar10-C and Cifar100-to-Cifar100-C you should train it yourself by setting --src_data_dir, source data Download

Source Model

• ImageNet-to-ImageNet-C: you can load the source model from timm.
• Cifar10-to-Cifar10-C: you can load the source model from here.
• Cifar100-to-Cifar100-C: you can load the source model from here.

Training

bash bash/imagenet.sh
bash bash/cifar10.sh
bash bash/cifar100.sh

Acknowledgements

Our code is based on the PyTorch implementation of the following projects:

• KFF
• CoTTA
• ViDA
• DPCore
• Robustbench