apchenstu / Facial_Details_Synthesis

[ICCV2019 Oral] Photo-Realistic Facial Details Synthesis from Single Image
MIT License
627 stars 131 forks source link

Cannot fit landmarks #46

Open ankarako opened 3 years ago

ankarako commented 3 years ago

Hello,

First of all, thanks for sharing your work. It seems that I cannot estimate the proxy mesh no matter the input image (even your paper's sample in Figure 1). Can you imagine what may be wrong? The log I'm getting is the following:

Load preTrain model done. Loading the model Model loaded Starting tracking "img_path" ===> Landmarks detection done.

===> estimating proxy of "img_path" ! ! Couldn't detect faces in the : "img_path"

Best regards, Antonis

apchenstu commented 3 years ago

did the problem solve? if not, please check the image path, it seems "img_path" is incorrect, you may need to replace the "img_path" with your real path, e.g., "e:/something/xxx.png"

XIAO-RUI-RUI commented 2 years ago

Maybe the picture's problem, some is ok, but some not.