Business Problem & Solution
BEL & eInfochips · Defence · 2016–2019 · Public
The problem
A secured site has more cameras than an operator can watch. Most of the time nothing happens; the rare event that matters — a person over the fence, a vehicle at a gate it shouldn't be at — is buried in hours of empty footage. Staring at a wall of monitors does not scale, and constant human attention is both expensive and unreliable.
The production system watches the feeds for the operator: it detects objects, tracks them, tests them against operator-drawn zones, and raises a governed alert only when something genuinely crosses a line — so the control room reacts to events, not to video.
The solution (downscaled)
This demo is a faithful, downscaled reimplementation on synthetic, iconographic perimeter frames — no real footage, no faces, no people. One vision stage reads each frame; everything else (filtering, tracking, zone tests, dwell confirmation, rules, governance, dispatch) is deterministic.
The honest cost story is built in: only the vision stage is metered, a change gate skips unchanged frames, and a dwell window kills single-frame false positives before any alert reaches an operator. In OSS mode the vision model runs on the secured network — nothing leaves the perimeter.
What it is not
No real cameras, no live video, no real alerting integration, no real site. Every frame is synthetic and labelled; the employers (BEL & eInfochips) are named only as context. The real detector, tracker and integrations are described in Architecture → Out of scope.