I believe this is the same project that surfaced a while back and the algorithm is remarkably effective for its simplicity, but I do agree that there are probably some others that would catch edge/sneaky cases.
The bigger issue are planes that are't showing up anywhere. I think it would be interesting to pair some audio and computer vision resources with existing ADSB network to train what planes sound like and possibly even which altitude/direction they are traveling based on sound energy alone.
Camera would be nice, also passive radar given that most of these ADSB tools are using existing software defined radios.