I haven't dug deeply, but there seems to be a whole ecosystem. An open source avatar data format[0], full body motion capture software[1], this stuff integrates with standard 3d software environments such as unity/unreal/blender/maya/max. So does commercial software[2]. For the hardware it can be a simple web/phone camera on the low end (giving you a talking head) over camera + VR trackers + an iphone for the face sensor to high end rigs with multiple cameras and mocap suits.

[0] https://vrm.dev/en/ [1] https://github.com/sh-akira/VirtualMotionCapture [2] https://www.youtube.com/watch?v=hJ_5d_1NHCQ

I don't think it's 100% the same thing (VRM appears to be full body, not just face), but FaceRig[1] is a similar concept made by a Romanian studio.

[1] https://facerig.com/