Audience analytics estimates who was actually in front of a screen: how many people, for how long, and sometimes broad age or gender brackets. It is the measurement that turns proof of play into something an advertiser can value.
The distinction from display analytics is the whole point. Display analytics says the asset played; audience analytics says somebody was there to see it. Only the second supports a comparison with other media.
The caveat is that most reported audience figures in out-of-home are modelled rather than counted. A number derived from traffic surveys, site type and assumed visibility is a projection, and two networks using different models produce numbers that cannot be compared even though both look like measurements. Ask which it is before treating a figure as observed.
Where sensors do count directly, the privacy line applies as it does everywhere else: aggregate counting is defensible, and estimating demographics from a camera edges toward biometric processing under GDPR and KVKK. Count at the device, keep nothing identifying, and be able to state in one sentence what is retained.