They can detect the presence of phones, yes. But that doesn’t automatically mean being able to detect people that aren’t carrying phones. To do that, you’d need to integrate the phone detection data with some other source of data on people present in the area in question. I’m saying there’s no evidence of such a system actually being used in practice. The paper you linked doesn’t address that at all.
Btw, to help understand the technical challenges involved with this, the whole reason Tesla focused on vision-only for its self-driving was the difficulty of integrating sensor data from multiple sources, e.g. lidar + vision would be significantly more difficult to achieve. It’s not that this isn’t possible in theory - it’s just that there’s no evidence of anyone having done it for “lack of phone” detection, and that’s probably because it’s not really a requirement that’s in high demand.
Btw, to help understand the technical challenges involved with this, the whole reason Tesla focused on vision-only for its self-driving was the difficulty of integrating sensor data from multiple sources, e.g. lidar + vision would be significantly more difficult to achieve. It’s not that this isn’t possible in theory - it’s just that there’s no evidence of anyone having done it for “lack of phone” detection, and that’s probably because it’s not really a requirement that’s in high demand.