Cameras

The camera is the first link in the recognition chain, and its choice shapes everything downstream — accuracy, latency, and hardware cost. Before selecting a model, you first decide where the video is processed. Smart Corridors supports two source types, and both are fully supported: classic IP cameras with server-side RTSP processing, and smart cameras or AI boxes that process video on the edge.

With server-side RTSP processing, an IP camera streams raw video to the server, where the Video Processing Platform performs face detection and matching. With on-edge processing, a smart camera or AI box runs detection — and optionally recognition — on the device itself, so the server handles only a fraction of the usual work. The two approaches differ mainly in where the compute happens and what that costs.

Both have trade-offs. RTSP remains very popular and stays fully supported for years to come, so it is a safe default for most deployments. On-edge processing is growing quickly because it lowers latency and reduces server-side hardware costs, which matters as a site scales to many cameras. Neither is strictly better — the right answer depends on your site size, network, and budget.

Read the two source-type pages to understand each approach, then use the selection guidance to pick hardware that performs well in the field:

• Server-side RTSP processing — classic IP cameras, all compute on the server.
• On-edge processing — smart cameras and AI boxes, compute on the device.
• Choosing a camera — practical selection criteria and a compatible-hardware list.