Awesome
Official Implement of ECCV 2024 paper "Reshaping the Online Data Buffering and Organizing Mechanism for Continual Test-Time Adaptation".
Preparation
Please create and activate the following conda envrionment.
# It may take several minutes for conda to solve the environment
conda update conda
conda env create -f environment.yml
conda activate ctta
We recommend referring to this repository to obtain the required datasets and source domain models. After downloading, please modify _C.DATA_DIR in conf.py accordingly.
Experiment Execution
Classification Experiments
You can use the provided configuration files to run experiments. Simply execute the following Python script with the corresponding configuration file:
# Tested on RTX4090
CUDA_VISIBLE_DEVICES=0 python test_time.py --cfg cfgs/[cifar10_c/cifar100_c/imagenet_c]/[source/norm_test/norm_alpha/tent/cotta/rotta/santa/obao].yaml
Alternatively, you can execute the provided shell script:
bash run.sh
Segmentation Experiments
We heavily rely on the Cotta code in segmentation experiments. Please follow the instructions provided in cotta to download the segmentation code and set up the environment.
Next, replace ./tools/our.py and ./mmseg/apis/test.py with the our.py and test.py files from the ./seg folder of this repository. Then, you can run the following command to perform the segmentation experiment:
CUDA_VISIBLE_DEVICES=0 python ./tools/our.py ./local_configs/segformer/B5/segformer.b5.1024x1024.acdc.160k.py segformer.b5.1024x1024.city.160k.pth --rounds 10
Citation
Please cite our work if you find it useful.
@inproceedings{zhu2024reshaping,
title={Reshaping the Online Data Buffering and Organizing Mechanism for Continual Test-Time Adaptation},
author={Zhu, Zhilin and Hong, Xiaopeng and Ma, Zhiheng and Zhuang, Weijun and Ma, Yaohui and Dai, Yong and Wang, Yaowei},
booktitle={European Conference on Computer Vision},
pages={415--433},
year={2024}
}