search

GilLevi / AgeGenderDeepLearning

628 stars 284 forks source link

How to affine the image? #1

Closed alfredtofu closed 6 years ago

alfredtofu commented 8 years ago

As mentioned in your paper, you use adience_align to align the faces. So can you briefly explain the alignment setting. Just like how big the face in the image.

GilLevi commented 8 years ago

Hi,

Thank you for your interest in our work.

Alignment is done by detecting about 68 facial landmark and applying an affine transformation to a predefined set of locations for those 68 landmarks.

I didn't understand what you mean by the size of the face in the image. Can you please elaborate?

Best, Gil

ma1onso commented 6 years ago

@GilLevi is needed alignment the images with https://github.com/eranid/adience_align before processing with your model? Thanks

GilLevi commented 6 years ago

Hi @AlonsoEnrique , if you want to process new images (not from the AdienceFaces set which are already aligned) then yes, you will get better results if you will align them first. You will get a bit worse results without alignemnt, but I can't say in advance how much the results would be worse.