How it works
Detect → decide → act
01 · Detect
Neural networks process live streams on the TRASSIR server. Objects, plates, faces and events surface in real time.
02 · Decide
Rules turn detections into alarms, reports and integrations, tuned per channel and per schedule.
03 · Act
Operators respond in the VMS; every event lands in the archive with evidence attached.
The modules
Five modules, measured in incidents prevented
TRASSIR Neuro Detector
Human and vehicle detection that ignores weather, animals and shadows.
Neural network object detector for security that accurately recognizes people, heads, cars, bicycles and other objects and raises an alar…
The moduleTRASSIR ActivePOS
Every receipt line synchronized with video, so you catch checkout fraud as it happens.
Cash desk analytics module that synchronizes POS events with video and runs each transaction through a violation checklist, generating an…
The moduleAutoTRASSIR (LPR/ANPR)
Automatic number plate recognition for gates, parking and logistics.
Intelligent license plate recognition module that automatically identifies plates on vehicles entering a camera's field of view and track…
The moduleTRASSIR Face Recognition
Watchlists, VIP recognition and access scenarios with photo evidence.
Face recognition module combining FaceSDK for face comparison with Face Search for database lookups. Detects faces from any angle, matche…
The moduleTRASSIR Hardhat Detector
PPE compliance monitoring on live video, straight from your cameras.
Work safety module that monitors compliance with protective helmet rules: when an employee appears in the work area without a hardhat it …
The moduleHow to buy
Pilot first, then scale what proves ROI
Modules are activated on TRASSIR servers you already run, with nothing to rip out. A typical engagement: we review camera placement against each module's requirements, pilot on selected channels, then scale what proves its return.
Server sizing matters: neural processing needs the right compute behind it. See the NeuroStation line under Servers
Choosing modules
Match the module to the problem
Every module produces the same currency: an event in the VMS with video evidence attached. The difference is what each neural network is trained to notice.
| The problem | The module | The event you get |
|---|---|---|
| Intruders on the perimeter, false alarms from weather and animals | Neuro Detector | An alarm only when a person or vehicle enters a drawn zone |
| Shrinkage and errors at the checkout | ActivePOS | An incident with a short synchronized video clip for every flagged transaction |
| Slow gates and unlogged vehicles | AutoTRASSIR | Every plate read and logged with snapshot and time; barriers open for authorized groups |
| Persons of interest walking in unnoticed | Face Recognition | A scored match with the photo pair, seconds after the face appears |
| Helmets coming off when nobody is watching | Hardhat Detector | A live notification with the frame saved as evidence |
Honest expectations
What analytics need from your system
Server compute
Neural processing runs on NeuroStation class servers, and offload analytics lets other TRASSIR servers send video to one machine for processing. We size this before quoting.
Image quality
A detector is only as good as the picture it sees. Placement, focus and lighting are reviewed per channel before any module is switched on.
Tuning on site
Zones, rules, schedules and thresholds are calibrated on your site, in your conditions. Accuracy is won in commissioning, not claimed in a brochure.
A measured pilot
We agree success criteria up front and run the modules on selected channels first, so scaling is a decision backed by your own results.
Questions we hear
Straight answers before you commit
Will the modules work with our existing cameras?
In most cases, yes. TRASSIR VMS supports 99% of cameras through native, RTSP and ONVIF, so analytics run on the streams you already record. We verify image quality and placement per channel.
Do we need to replace our recording servers?
Not necessarily. Neural modules run on NeuroStation servers, and offload analytics lets multiple TRASSIR servers send video to one machine, so existing recorders keep their role.
Can we start with just one module?
Yes. Modules are activated on the channels where they earn their keep. The usual path is a pilot on selected cameras, then scaling what proves its return.
How do operators receive events?
Inside the VMS they already use: alarms, archived evidence, smart search and reports, plus iOS and Android apps with push, Telegram and Discord notifications.
How accurate are the detectors?
Accuracy depends on placement, light and tuning, so we prove it on your own footage in a pilot instead of quoting universal percentages.
Pilot AI analytics on your existing cameras
Send your camera layout and the problems you want solved. We'll match the modules, size the compute and scope a pilot you can measure.