01
Quality and material inspection
Parts, tools or products inspected on the line; cracks, chips, surface anomalies or spills flagged for human intervention.
Flexible computer vision
AI Vision
Identify, localise and highlight any user-defined element.
A trainable computer-vision foundation. Instead of predefined objects or tasks, you train it with your own images and labels so it recognises, locates and highlights what matters — from living subjects and movement to material defects — with no hard-coded updates.
- learned from your own images and labels
- Any class
- deployment, containerised
- Edge or cloud
- to deploy
- Ready
A moment with AI Vision
A technician labels forty photos of a hairline crack. By the afternoon the line camera reports "DEFECT · 98.2% · surface_anomaly" on a part the human inspector would have passed at that speed, and the part is pulled before it reaches the carton.
The problem it removes
Off-the-shelf vision systems recognise what their vendor trained them on.
The crack on your part, the fixture on your pole, the pallet in your lane are not on the list, and a new class means a new project. The answer is a foundation that learns from your samples.
01 / 06Capability 1
Open-ended classification
You set the rules. The platform learns from your images and labels to find objects, detect living subjects, count items or flag custom conditions.
02 / 06Capability 2
Dual-purpose tracking
Spatial awareness — presence, foot traffic, occupancy — and fine-grained quality control — surface wear, tears, structural anomalies — in one engine.
03 / 06Capability 3
Multi-format processing
Still photos, short clips and continuous surveillance streams, analysed with the same models.
04 / 06Capability 4
Hardware and angle agnostic
Works across capture angles, distances, lighting and camera models, from mobile devices to fixed industrial lenses.
05 / 06Capability 5
Edge and local deployment
Runs in the cloud or in lightweight containerised modules on on-site edge devices for lower latency and isolated data.
06 / 06Capability 6
Visual alerts for people
Flags cracks, chips, spills, missing fixtures or structural shifts with a highlighted frame, so a human decides fast.
How it works
Sense. Understand. Decide. Act.
- 01SenseCameras, radar, microphones, environmental and access sensors — chosen for the space, not the brochure.
- 02UnderstandVision and speech models run on an edge gateway in the room, so nothing waits for the cloud and nothing leaves it without consent.
- 03DecideScene rules and safety policies turn detections into decisions: escalate, notify, actuate, or stay quiet.
- 04ActLights, locks, motors, doors, screens, speakers and people — through open protocols and one operations console.
Operations
One console. Every event, with its evidence.
Live status, detections with the clip or image that triggered them, who was notified and what happened next. Per room, per car, per gate, per site.
| Training | Custom images and labels; learning engine trains per deployment |
|---|---|
| Inputs | Photos, clips, live streams; any camera, any angle |
| Outputs | Classes, locations, counts, highlighted frames, alerts |
| Deployment | Cloud or on-site edge containers |
| Tracks | Spatial and flow awareness; quality and inspection audits |
Integrations
Deployments
Where it goes to work.
02
Spatial and flow auditing
Occupancy, asset counts and item movement mapped in warehouses, public hubs or logistics bays.
03
Infrastructure health monitoring
Cameras on field gear or mobile platforms detect missing fixtures, structural shifts or surface wear across large areas.
Questions
What clients ask first.
How many sample images do we need?
It depends on the class, but a few dozen labelled examples per condition is a normal starting point; the model keeps improving from operator feedback.
Does video leave our site?
Not if you deploy at the edge. The containerised modules run on on-site hardware and only the results you choose are shared.
Is this the engine behind Argus Vision?
Argus Vision is the deployed safety product; AI Vision is the trainable foundation it builds on, available for your own classes.
NT-03 · AI Vision
Tell us about your site. We come back with a design and a written quote.
A floor plan, deck plan, yard layout or garden drawing is the best start. Hardware is sourced from factories Outsource China inspects, and delivered on one contract.
Related
Works well with
EG-04Safety & securityArgus VisionCameras that understand what they see: fire, intruders, a fall in the hallway.Explore →
EG-06ManufacturingForge FloorA factory floor that sees every risk and every defect — in real time.Explore →NT-02Operational performance analyticsAI ProductionMonitor output, track targets and ask in plain language.Explore →