Multi-modal Face Anti-spoofing (Presentation Attack Detection) Challenge@CVPR2019
How to participate this challenge and dataset download. It includes three steps below:
Register for the Chalearn Face Spoofing Attack Detection Challenge (codaLab website) [challenge is finished, please ignore this step].
2. Download, read the Contest Rules of the CASIA-SURF dataset, and sign the agreement (link).
3. Send the your signed agreements to: Jun Wan, jun.wan@ia.ac.cn
Notes: If you tend to get the commercial license of this dataset, please email to contact@surfingtech.cn
Q&A
Q: Fine-grained spoof labels
A: For the CASIA-SURF dataset, the attack material is paper, including 6 cutting ways. In which the ‘4’, ‘5’ and ‘6’ are used for training, and ways ‘1’, ‘2’ and ‘3’ are used for testing. In the dataset name, "04_en_b.rssdk" means that the cutting method is ‘4’, and other ways are similar.