FashionAdv: Fashion-Guided Adversarial Attack

Setup

To reproduce the paper results you will need to download few things.

The weight for yolact

• Can be found here: https://drive.google.com/uc?id=1UYy3dMapbH1BnmtZU4WH1zbYgOzzHHf_&export=download

The COCO 2017 validation dataset.

• wget http://images.cocodataset.org/zips/val2017.zip -O ./data/val2017.zip

• unzip data/val2017.zip

• rm data/val2017.zip

• wget http://images.cocodataset.org/annotations/annotations_trainval2017.zip -O ./data2/annotations_trainval2017.zip

• unzip data/annotations_trainval2017.zip

• rm data/annotations_trainval2017.zip

Pre-computed data

• unzip data/Archive.zip

Execute the attack

Path for data is given in the configuration.json file.

To Execute the attack you can use:

python3 run_fashionAdv_attack.py

It will run FashionAdv on same setup and dataset as in the paper.

You can modify value in the configuration.json file.

You can also run the code on a smaller part on the dataset by given 2 argument to the script

FashionAdv_attack.py id_first_image id_last_image

python3 run_fashionAdv_attack.py 0 100

Evaluate the attack

python3 evaluate_fashionAdv_attack.py 0 100

Citations

Please consider citing this project in your publications if it helps your research:

@Inproceedings{marc-CVPRW2021,
  Title          = {Fashion-Guided Adversarial Attack on Person Segmentation},
  Author         = {Marc Treu and Trung-Nghia Le and Huy H. Nguyen and Junichi Yamagishi and Isao Echizen},
  BookTitle      = {Conference on Computer Vision and Pattern Recognition Workshops},
  Year           = {2021},
}

License

The code is used for academic purpose only.

Contact

Marc Treu, Trung-Nghia Le.