In fact it would be ”one order of magnitude faster” in fancy talk, as it could process in under 5ms, or 200 times a second, all thanks to ‘3D depth maps’.
Basically they analyse a load of ‘3D maps’ of people and then aggregated the data letting a new AI routine be created that can much more easily identify body parts across a much wider selection of users. This speeds up the tracking process a lot.
Below is a video from Microsoft Research.

Loading comments…