clks-wzz / FAS-SGTD

Deep Spatial Gradient and Temporal Depth Learning for Face Anti-spoofing

FAS-SGTD

---

Introduction

Main codes of CVPR2020 paper "Deep Spatial Gradient and Temporal Depth Learning for Face Anti-spoofing"

Prerequisite

• Python 3.6 (numpy, skimage, scipy)

• TensorFlow >= 1.4

• opencv2

• Pillow (PIL)

• easydict

Train

You can use PRNet to generate the virtual depth maps.

For single-frame stage:

cd fas_sgtd_single_frame && bash train.sh

For multi-frame stage:

cd fas_sgtd_multi_frame && bash train.sh

Test

We provide an example for Protocol 1 of OULU-NPU. You can download the models at BaiduDrive(pwd: luik) or GoogleDrive and put them into fas_sgtd_multi_frame/model_save/.

cd fas_sgtd_multi_frame && bash test.sh

License

Code: under MIT license.

It is just for research purpose, and commercial use is not allowed

Citation

If you use this code, please consider citing:

@inproceedings{wang2020deep,
    title={Deep Spatial Gradient and Temporal Depth Learning for Face Anti-spoofing},
    author={Wang, Zezheng and Yu, Zitong and Zhao, Chenxu and Zhu, Xiangyu and Qin, Yunxiao and Zhou, Qiusheng and Zhou, Feng and Lei, Zhen},
    booktitle= {CVPR},
    year = {2020}
}

@inproceedings{yu2020searching,
    title={Searching Central Difference Convolutional Networks for Face Anti-Spoofing},
    author={Yu, Zitong and Zhao, Chenxu and Wang, Zezheng and Qin, Yunxiao and Su, Zhuo and Li, Xiaobai and Zhou, Feng and Zhao, Guoying},
    booktitle= {CVPR},
    year = {2020}
}

@inproceedings{qin2019learning,
    title={Learning Meta Model for Zero-and Few-shot Face Anti-spoofing},
    author={Qin, Yunxiao and Zhao, Chenxu and Zhu, Xiangyu and Wang, Zezheng and Yu, Zitong and Fu, Tianyu and Zhou, Feng and Shi, Jingping and Lei, Zhen},
    booktitle= {AAAI},
    year = {2020}
}