Video Analytics and AI-Assisted Detection

Analytics is bought on the demonstration and judged on the false positive rate.

An IVA demonstration runs on reference footage, in controlled light, at an ideal camera height. Your car park at 6 p.m. in monsoon rain is a different problem, and only a figure measured there will predict how it performs.

Talk to SECCOM about running a real accuracy baseline on your own footage for two weeks. Detections, misses and false positives, counted per camera.

SECCOM writes the decision boundary into the design. The system flags, ranks and routes. A person decides. Nothing in a SECCOM design takes an action against an individual on its own.

What you get. Per-camera detection and false-positive counts from your site, a camera placement and lighting change list, usually the cheapest accuracy improvement available, and a written statement of what the system may do unattended and what requires an operator.

Many estates already run Dell compute and storage for retention and analytics. The baseline runs on it without changing anything already in place.

Where SECCOM stops. SECCOM does not deploy analytics that act automatically against a person, and reports only accuracy measured on your own estate.