elroborn / dfa Goto Github PK

thanks for your great contributions on face anti-spoofing, i wonder how many frames do you get from each video on oulu-npu datasets, thanks

thanks,your aritcle is very good.when I run your code,I have a question,where can we get the "/data/oulu/depth/Train_files/1_1_01_1_avi/frame4.jpg" dataset.

Hi. Your idea is really nice. But i have a doubt what are those latent variables(z) and stacked features(response map z')? And, When do you use the classifier? after training the GAN?.If so do you store the latent variables to give them to the classifier. What do Model1, Model2,Model3 represent?

In the paper,I can't understand the "the ground truth depth image" amd how to get the "the ground truth depth image",can you explain? thanks!