search

SoftwareGift / FeatherNets_Face-Anti-spoofing-Attack-Detection-Challenge-CVPR2019

Code for 3rd Place Solution in Face Anti-spoofing Attack Detection Challenge @ CVPR2019,model only 0.35M!!! 1.88ms(CPU)
Other
928 stars 284 forks source link

Real-time Demo #60

Closed jeffryuiop closed 4 years ago

jeffryuiop commented 4 years ago

In your demo (https://www.youtube.com/watch?v=giznA4WU2-4), which model did you use? Was it the ensemble, or was it only FeatherNetB?

SoftwareGift commented 4 years ago

In your demo (https://www.youtube.com/watch?v=giznA4WU2-4), which model did you use? Was it the ensemble, or was it only FeatherNetB?

Only FeatherNetB

jeffryuiop commented 4 years ago

okay thank you for the reply. I tried to make a demo using your pretrained weight, but the result is not as good as yours (youtube video). To be honest I'm still confused with the inference process that you stated in other forum. It would be great if you can clarify a few things.

  1. So the output of the model is 1024 vector, but you only use the first 2 to put into softmax in order to make the inference?
  2. Did you also use FeatherNetB for IR in the demo?