xunguangwang / pros-gan

[CVPR 2021] Official repository for "Prototype-supervised Adversarial Network for Targeted Attack of Deep Hashing"

Prototype-supervised Adversarial Network for Targeted Attack of Deep Hashing

This is the code for our CVPR 2021 paper "Prototype-supervised Adversarial Network for Targeted Attack of Deep Hashing", which formulates a flexible generative architecture for efficient and effective targeted hashing attack. In this repository, we not only provide the implementation of the proposed Prototype-supervised Adversarial Network (i.e., ProS-GAN), but also collect some popular deep hashing methods used in the paper and the previous targeted attack methods in hashing based retrieval.

Usage

Dependencies

• Python 3.7.6
• Pytorch 1.6.0
• Numpy 1.18.5
• Pillow 7.1.2
• CUDA 10.2

Train hashing models

Initialize the hyper-parameters in hashing.py following the paper, and then run

python hashing.py

Attack by P2P or DHTA

Initialize the hyper-parameters in dhta.py following the paper, and then run

python dhta.py

Train ProS-GAN

Initialize the hyper-parameters in main.py following the paper, and then run

python main.py --train True

Evaluate ProS-GAN

Initialize the hyper-parameters in main.py following the paper, and then run

python main.py --train False --test True

Cite

If you find this work is useful, please cite the following:

@inproceedings{wang2021prototype,
	title={Prototype-supervised Adversarial Network for Targeted Attack of Deep Hashing},
	author={Wang, Xunguang and Zhang, Zheng and Wu, Baoyuan and Shen, Fumin and Lu, Guangming},
	booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
	year={2021}
}

pros-gan's Issues

You are quite outstanding！

You are quite outstanding！

Pre-trained Models

Hi,
This is a great job!
Can you release your pre-trained models?
Thanks!

Nice work!

Nice work!

FileNotFoundError: [Errno 2] No such file or directory: './data/NUS-WIDE\\images/0041_184606467.jpg'

After running the hashing code, the file data/NUS-WIDE\images/0041_184606467.jpg cannot be found

No such file or directory: './checkpoint/ImageNet_DPSH_VGG11_32.pth'

hello,
When running main.py, there is a model-free error. The two models I saw in checkpoint are not the same as the error. I don't know if I can use them. What should I do? Come to consult, hope to get your answer, thank you very much.