mkorunoski / 3d-face-reconstruction

3D reconstruction of detected face

3D Face Reconstruction

3D reconstruction of detected face. dlib and OpenCV libraries are used for face detection and image manipulation. OpenGL is used for rendering.

Reference materials

• [Head pose] (http://stackoverflow.com/questions/36590516/how-to-get-3d-coordinate-axes-of-head-pose-estimation-in-dlib-c)
• [Head pose revisited] (http://www.morethantechnical.com/2012/10/17/head-pose-estimation-with-opencv-opengl-revisited-w-code/)
• [Speeding up Dlib’s Facial Landmark Detector] (http://www.learnopencv.com/speeding-up-dlib-facial-landmark-detector/)

Screenshots

• Face detection

• Feature points extraction

• Face model for mapping

• Final result

3d-face-reconstruction's Issues

Male pictures

Hi,
a fantastic project.
I tried to use the obama picture, in the following release, but I got a distored picture.

Also another bug, when rotating the model, the model and texture doesn't rotate too.

3D Reconstruction with different images

3D Face Reconstruction from an image successfully is made by your application with the images provided in application. But I have problems when trying to reconstruct with different images. I just read
the blog post which you linked in the comment section http://www.morethantechnical.com/2012/10/17/head-pose-estimation-with-opencv-opengl-revisited-w-code/. In the blog post; some 3d and 2d points related to Left Eye, Right Eye, Left Ear, Right Ear, Left Mouth, Right Mouth and Nose are set manually. So I think this is the reason of having problems with reconstruction with different images.Do you have any suggestion about dynamically find these position?