Draw the line.Alert when it is crossed.
Fenec detects people, forklifts, and vehicles entering restricted zones, forklift lanes, or maintenance areas at the wrong time.
Danger often happens before the collision.
Near-misses are only useful if the site can see and act on them.
Rules that understand space, class, and time.
The same zone can allow forklifts, reject pedestrians, and change behavior during maintenance.
Polygon zones
Define exact areas directly on each camera view.
Class-specific logic
Allow vehicles while alerting on pedestrians, or the reverse.
Low-latency inference
Local processing keeps alert latency practical.
Escalation rules
Route incidents by severity, shift, and destination system.
Designed for fast, reviewable alerts.
Pilot thresholds should be tuned against your camera angles and traffic patterns.
Write rules operators can reason about.
Zone alerts should be explicit enough to audit and simple enough to maintain.
when: person_in_zone zone: forklift_lane dwell_ms: 1200 notify: supervisor
when: vehicle_in_zone zone: maintenance_area schedule: closed_hours notify: security
when: forklift_direction zone: loading_lane direction: against_flow notify: operations
Zone model configuration
Reference configuration for pilot scoping. Final thresholds are tuned to each site.
| Inputs | RTSP / ONVIF video streams |
| Classes | Person, forklift, vehicle |
| Rules | Zone, direction, dwell, schedule |
| Outputs | Metadata, alert, optional redacted clip |
Zone event flow
Structured events are emitted to your systems while raw video remains on the appliance.
Estimate time saved from manual checks.
Use this as a scoping tool, not a guaranteed financial claim.
Formula
Final ROI depends on site workflow, escalation path, and actual incident reduction.
Turn near-misses into usable signals.
Scope a high-risk lane or restricted area and measure whether alerts are accurate enough for operators to trust.
Local processing keeps the privacy review narrow.