how to get translation_z about face-pose-net HOT 6 OPEN

Hi, I'm doing the same job, so how about your works?
Did you output the tz and what is the mean of scale?
Thank U!

from face-pose-net.

I am also curious about getting tz from the scale on rigid transformation.

from face-pose-net.

Hey all, I am not sure how well this repository is still being maintained.
I can suggest, however, that you instead look at our follow-up project, img2pose, which provides the same 6DoF output for each face in the image, but does not require face detection: It uses a single, small model.
I believe you'll find that project also better supported.

from face-pose-net.

Hey all, I am not sure how well this repository is still being maintained. I can suggest, however, that you instead look at our follow-up project, img2pose, which provides the same 6DoF output for each face in the image, but does not require face detection: It uses a single, small model. I believe you'll find that project also better supported.

Hi, img2pose is an awesome project. However, I tried then It seemd the translation z have a large error when i used web camera to test the head 6DOF pose.I don't know if the camera intrinsic or parameters should be adjusted.
Looking forward to your reply!

from face-pose-net.

Hi, I knew that camera intrinsic parameters could be printed using this: http://www.vision.caltech.edu/bouguetj/calib_doc/

from face-pose-net.

Thanks, @eugeneYz , really appreciate it! But I think the best person to answer your question is probably Vitor, who's maintaining the img2pose code repo. I suggest posting the question there.

from face-pose-net.