Why it exists
It recognizes what is moving, not just that something moved
Classic motion detection alarms on everything that moves: rain, headlights, stray dogs, branches in the wind. Neuro Detector replaces pixel change logic with a trained network that recognizes what is moving, so operators see genuine human and vehicle activity instead of drowning in false alarms.
Because it runs inside TRASSIR VMS, every detection is an ordinary alarm event: it triggers recording, notifies operators, appears in smart search and feeds reports, with no extra software layer.
Demo footage: official TRASSIR product page
Capabilities
What the module does, as TRASSIR publishes it
- Highly accurate object detection and classification (people, heads, cars, bicycles, etc.)
- Alarm generation when foreign objects enter specified areas
- Improved control over territory and access to sterile areas
- Sharply reduced false alarms: unlike motion detectors, it ignores trees, animals and noise
- Timely response to sensitive events on site
- Offload analytics: multiple TRASSIR servers send video to one server for processing
What it solves
Fewer, cleaner alarms and a perimeter you can trust
Perimeter protection
Alert only when a person or vehicle crosses the line, not when the weather does.
Monitoring after hours
Arm zones by schedule; your guard force responds to verified events instead of noise.
Operator efficiency
Fewer, cleaner alarms mean a small control room team can cover more cameras.
Chasing Sun engineers calibrate zones, sensitivity and schedules per site during commissioning; accuracy is won in the tuning, not the brochure.
Inside the module
From pixels to a verified alarm
The neural network classifies what it sees in each stream: people, heads, cars, bicycles and other trained object types. You draw detection zones per channel and choose which classes matter there. When a watched class enters a zone, the module raises an alarm event in TRASSIR VMS: recording triggers, operators are notified, and the event is archived with footage for smart search and reports.
The operator never has to interpret raw motion. Each alarm arrives already classified, with the zone and the object type that caused it.
What a deployment needs
- Compute: the module runs on TRASSIR NeuroStation servers, and other TRASSIR servers can offload their video to one NeuroStation for processing, so existing recorders keep their role.
- Camera placement: a stable mount with a clear, unobstructed view of the zone; the objects you care about must appear large enough in the frame to classify.
- Lighting: the network needs a usable image, so plan visible light or IR illumination for zones that matter at night.
| Use it for | The zone you draw | The alarm you get |
|---|---|---|
| Perimeter after hours | The fence line and the approach strip | A person or vehicle entering while the zone is armed |
| Sterile and restricted areas | Server rooms, stock cages, plant rooms | Any person entering a zone where nobody should be |
| Yard and vehicle control | Loading yards and internal roads | A vehicle appearing where, or when, it should not |
From pilot to production
Prove it, tune it, then scale it
01 · Demo
See the module live on demo footage, or run a proof of concept on recordings from your own site.
02 · Scoped pilot
We activate the module on selected channels of your cameras and agree what a successful result looks like.
03 · Tuning
Zones, object classes, sensitivity and schedules are calibrated on site until the alarms are clean.
04 · Rollout
Scale to the full camera estate, with operator training and support from our Nairobi team.
Questions we hear
Straight answers before you commit
Do I need to replace my cameras?
No. The module analyzes video from the IP cameras you already run in TRASSIR VMS. What matters is image quality: a sharp, well placed view of the zone. We check each camera during scoping.
Where does the processing happen?
On a TRASSIR NeuroStation server. Other TRASSIR servers on the network can offload their analytics to one NeuroStation, so you concentrate the neural compute instead of upgrading every recorder.
How is this different from motion detection?
Motion detection reacts to pixel change, so rain, headlights, animals and branches all raise alarms. Neuro Detector classifies the object first and alarms only on the classes you choose.
Does it work at night?
It works wherever the camera produces a usable image. For zones that matter after dark, we plan visible or IR illumination as part of the design.
How accurate is it?
Accuracy depends on camera placement, image quality and tuning, which is why we commission it on site and prove it in a pilot on your own footage rather than quoting a brochure figure.
Prove it on your own footage
Pilot Neuro Detector on channels of your existing TRASSIR system, and measure the alarms it raises and the noise it removes before scaling.