I remember reading this paper and I think watching a corresponding PoC video when this came out. Does anyone remember why or happen to know why this never made it to market? Error rate, maybe?

It already is in the market, just that it is not usually sold as a standalone product like Leap Motion. Through wall imaging using WiFi imaging is old news. Also, for gesture detection specifically, radar chips are often preferred over hacking WiFi firmware/additional signal processing.

I often see awesome papers but don’t see any way to replicate it short of coding the thing from scratch (if it actually worked). I think many of these are concepts that work in very controlled circumstances where the inference setup is probably well calibrated for the setting. I’d venture getting a more general purpose library or tool would be far more work.

I think the paper is just a research project. Considering that you need a wifi on, it would not work very well on the go.

But I am pretty sure there are devices on the market just not for the use case presented here.

There are a few papers that look at the applications of what you can do with radio waves as a way to detect movements.

Very much not my area, but would this also mean that there is a way to parse the same signals they used for identifying finger movements to build a more general 3D map of the room? Perhaps with variances between multiple WiFi signals if you know the origin?

There's been a few studies on this. Some for positioning, others for slightly less comfortable things like this [0].

[0] http://people.csail.mit.edu/fadel/wivi/