search

MarekKowalski / DeepAlignmentNetwork

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

how compute the meanshape s0 ? #52

Open stone226 opened 5 years ago

stone226 commented 5 years ago

hi,thanks for your work. I am confused about s0. how compute the meanshape s0 ?can you share the code about it.Dose it related to the face box? hope your reply

MarekKowalski commented 5 years ago

Hi,

I used the method described in Appendix A of this old paper. Unfortunately this was a while ago and I no longer have the code.

Best regards,

Marek

stone226 commented 5 years ago

hi,thanks for your answers. I have another problem. now,I have trained a model just using the stage1,and I found that if use the public data with its face boxes provided,the result is good.but if use same picture and opencv face detector,the result is bad. I doubt that the result is sensitive to the face detector if just using the stage1. can you give some suggesions ?

MarekKowalski commented 5 years ago

What settings are you using for data augmentation, are they the default ones? If so, it should work fine and I would suspect a bug.

ming9512love commented 3 years ago

hi, I am also confused about s0. If you have solved this problem can you share the code for meanshape of trainset? thanks

hi,thanks for your answers. I have another problem. now,I have trained a model just using the stage1,and I found that if use the public data with its face boxes provided,the result is good.but if use same picture and opencv face detector,the result is bad. I doubt that the result is sensitive to the face detector if just using the stage1. can you give some suggesions ?