[Corazoor]

Probably by learning how humans have to balance their extremities to remain standing upright. The camera angle with respect to the enviroment is likely a factor too, since errors got larger when outdoors.

Errors are also not exactly small, ~6cm average, although it's more like 1-2 or 5-12, depending on body part. I think this would very likely be noticeable in VR applications, but it is still very impressive accuracy overall.

The question also nicely highlights the disadvantages of trained algorithms: No one knows for sure, and it certainly isn't obvious by looking at the network weights...

On the other hand, when looking at the pictures, I felt like the authors: There should be enough information in there to get at least a good estimate. And it is extremely useful that one can nowadays "just" train a model to confirm such theories.

[idiotsecant]

Has the result been reproduced? I could quite easily move my forearm to several distinct positions not visible from that view that would have no impact on the rest of my posture. I feel like there's a lot of ML out there making big claims and not distributing the models for verification.