FeatherNets_Face-Anti-spoofing-Attack-Detection-Challenge-CVPR2019 icon indicating copy to clipboard operation
FeatherNets_Face-Anti-spoofing-Attack-Detection-Challenge-CVPR2019 copied to clipboard

Published 20 hours ago • verified publisher icon SoftwareGift

Have you tried training the RGB dataset with FeatherNet?

Open Center0710 opened this issue 6 years ago • 5 comments

Because my data set is only RGB, But I use you provide model(FeatherB) to do inferences very poorly (I know you use depth-image to training)

So I want to know if you have used RGB for training ? Can you share the results and let me know? Or you have suggested methods ? Thank you.

Center0710 avatar Oct 03 '19 04:10 Center0710

Because my data set is only RGB, But I use you provide model(FeatherB) to do inferences very poorly (I know you use depth-image to training)

So I want to know if you have used RGB for training ? Can you share the results and let me know? Or you have suggested methods ? Thank you.

+1

xiezheng-cs avatar Oct 05 '19 08:10 xiezheng-cs

The same question.

paleomoon avatar Feb 13 '20 03:02 paleomoon

+1

hustzeyu avatar Jun 19 '20 08:06 hustzeyu

你们可以自测一下啊。我现在用了depth，和ir图像，depth基本可以到99%，ir最高到87%并且还是很容易过拟合的。rgb的效果，也可以试一下，但是应该是比ir还要差的。

riskeverything avatar Oct 14 '20 02:10 riskeverything

请问大家有ir数据集吗？

promise12344321 avatar Nov 05 '20 07:11 promise12344321