I am searching a way to recalculate the current Position of the Face in "Virtual World Position" From a Webcam Texture. I was thinking about a way to solve this via taking the Average Pupillary distance (which is ~60mm). And... With this number I somehow need to get a actual position ... somehow... of the... Face... (The Center)

I dont know how exactly though. Any Ideas?

Explaining Image

  • \$\begingroup\$ Are you using any existing computer vision libraries, or trying to roll your own solution from scratch? Having done this myself, I don't recommend it — it's very fiddly, and difficult to match the tracking quality of established solutions. As a starting point, have you read up on face detection and tracking methods and decided which ones you plan to use? \$\endgroup\$ – DMGregory May 10 '17 at 13:58
  • \$\begingroup\$ At the Moment im at very the beginning of my journey. I thought of getting the coordinates of my eyes from existing assetstore-assets or some Framework. Im not sure there is any need for doing it from scratch right now. Is there any "easy-to-use-thing" ? I only need the coodinates of the eyes. \$\endgroup\$ – OC_RaizW May 10 '17 at 14:13
  • \$\begingroup\$ My Question here was not for getting ideas for a framework, but how to convert the data from the Textures to Virtual-World-Positions. \$\endgroup\$ – OC_RaizW May 10 '17 at 14:28
  • \$\begingroup\$ Ah, so you have the 2D coordinates of the eye pixels in image space, and you need the 3D coordinates in a camera-relative real world coordinate space? Do you have access to parameters describing the webcam's projection (field of view / focal length, lens distortion, etc)? \$\endgroup\$ – DMGregory May 10 '17 at 14:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.