search

MarekKowalski / DeepAlignmentNetwork

A deep neural network for face alignment
MIT License
509 stars 138 forks source link

About input image's size and the fc layer? #13

Open hangkaitan opened 6 years ago

hangkaitan commented 6 years ago

Hi, Do you explore differents input image's size? and can we use convolution layers instead of fc layer at the end of the net work? Or the fc layer has some advantage in alignments?

MarekKowalski commented 6 years ago

Hi,

I don't think I explored those two things. The size of the input and the fc layer were both based on the VGG network design. I imagine that a larger input image might improve accuracy, I don't think removing the fc layer would give an improvement, but it's never bad to give it a shot.

Best regards,