On-device processing
On-device processing is the architectural principle of running all AI inference locally on the in-store APU, so that raw video never leaves the building.
On-device processing is the architectural principle of running all AI inference locally on the in-store APU, so that raw video never leaves the building. It is the foundation of Aura Vision’s privacy guarantee — because no footage is transmitted or stored centrally, there is no central pool of imagery that could be accessed, leaked or misused.
How it works
Existing IP, RTSP or ONVIF cameras stream to the APU over the local store network. The APU runs the computer-vision models in real time: each frame is analysed in milliseconds and then discarded. Only the resulting anonymous, aggregated counts — visitor numbers, dwell times, demographic bands, queue metrics — are encrypted and sent to the cloud, at under 50 Kb/s per camera.
The result is that personal data is never produced in the first place. Privacy is a property of the architecture, not a policy bolted on afterwards.