Paper code: โEfficient Decision-based Black-box Patch Attacks on Video Recognitionโ.
git clone https://github.com/deep-kaixun/STDE.git
cd STDE
conda env create -f environment.yaml
C3D are from 3D-ResNets-PyTorch. NL and TPN models are from GluonCV
Download here.
python3 attack.py --batch_size 1 --cf_rate 0.6 --time_mua 1
python3 attack.py --batch_size 1 --cf_rate 0.7 --time_mua 2 --targeted
If you use the code or find this project helpful, please consider citing our paper.
@article{jiang2023efficient,
title={Efficient Decision-based Black-box Patch Attacks on Video Recognition},
author={Jiang, Kaixun and Chen, Zhaoyu and Huang, Hao and Wang, Jiafeng and Yang, Dingkang and Li, Bo and Wang, Yan and Zhang, Wenqiang},
journal={arXiv preprint arXiv:2303.11917},
year={2023}
}