Official PyTorch github repository for the paper Low-budget Label Query through Domain Alignment Enforcement published in the Computer Vision and Image Understanding (CVIU) journal, 2022.
- Pytorch 1.4.0
- Python 3.5
- Install the required packages:
pip install -r requirements.txt
- CIFAR-10 -> STL:
./digits.sh cifar9 stl9 dialnet
- STL -> CIFAR-10:
./digits.sh stl cifar9 dialnet
- SVHN -> MNIST:
./digits.sh svhn mnist dialnet
- MNIST -> USPS:
./digits.sh mnist usps dialnet
- USPS -> MNIST:
./digits.sh usps mnist dialnet
- ImageNet -> STL:
./imagenet.sh stl9 resnet50
- ImageNet -> CIFAR-10:
./imagenet.sh cifar9 resnet50
- ImageNet -> MNIST:
./imagenet.sh mnist resnet50
- ImageNet -> USPS:
./imagenet.sh usps resnet50
- ImageNet -> SVHN:
./imagenet.sh svhn resnet50
- Office-31: To run the experiments on the Office-31 dataset first you need to download the dataset from this page.
for SOURCE in amazon dslr webcam
do
for TARGET in amazon dslr webcam
do
if [[ "$SOURCE" != "$TARGET" ]]
then
./office-31.sh "$SOURCE" "$TARGET" resnet50
fi
done
done
- Office-Home: To run the experiments on the OfficeHome dataset first you need to download the dataset from this page.
for SOURCE in Art Clipart Product RealWorld
do
for TARGET in Art Clipart Product RealWorld
do
if [[ "$SOURCE" != "$TARGET" ]]
then
./office-home.sh "$SOURCE" "$TARGET" resnet50
fi
done
done
If you find this code useful for your research, please cite our paper:
@Article{journals-cviu-saltori-rsa-22,
author = {Cristiano Saltori and
Paolo Rota and
Nicu Sebe and
Jurandy Almeida},
doi = {10.1016/j.cviu.2022.103485},
journal = {Computer Vision and Image Understanding},
pages = {103485},
title = {Low-budget Label Query through Domain Alignment Enforcement},
volume = {222},
year = {2022},
publisher = {Springer}
}