CV use-case for raw camera access #3
Description
My research group at CMU has been experimenting with WebXR and definitely agrees that camera access for CV operations would enable a bunch of interesting use-cases. Being able to perform tag tracking is an obvious one (kind of step backward from AR.js at this point unless the two can coexist nicely and share the camera). We are experimenting with new types of markers like Light Anchors (https://karan-ahuja.com/assets/docs/paper/lightanchors.pdf) that won't be supported in AR Kit / Core anytime soon. Newer versions of the light anchor provide challenge response authentication which could be useful for certain applications. There is a never ending amount of scene perception and object recognition tasks that people would surely add.