Chalearn Single-modal (RGB) Cross-ethnicity Face anti-spoofing Recognition Challenge@CVPR2020 Forum

Go back to competition Back to thread list Post in this thread

> Question from participants

1. Can we use the sub-protocol "dev (w/ given label)" and "test (w/o label)" for the sub-protocol model training?
---No, you cannot use the label of development subset and test subset in the training stage. In the training stage, you can only use the provided training data.

3. Public ImageNet pretrained model can used for model initialization?
--- No, you cannot use the pre-trained model from ImageNet or other datasets.

4. Some non-deep learning depth estimation model, e.g. BFM(https://faces.dmi.unibas.ch/bfm/), can be used for single RGB track?
--- Yes, It's no problem. You can select non-DL methods.

5. Is there a WeChat group or QQ group for the challenge?
---I have createed a group via wechat. You can add me wechat: 13810491593.

Posted by: gesture_challenge @ Feb. 19, 2020, 5:24 a.m.
Post in this thread