Vision AI Safety Platform
QVision
Vision AI Safety Platform
QVision
Existing CCTV analysed live — detecting incidents and unsafe behaviour, masking personal data on the spot.
Overview
Cameras you already have, working as traffic and safety sensors
CCTV footage gives traffic counts by vehicle type, and potholes and incidents are detected live and passed to the control room.
Traffic flow
Counts by vehicle type, congestion and changes in speed, hour by hour
Spotting danger
Wrong-way driving, stopped vehicles, potholes and unsafe behaviour — caught before they become accidents
Into the control room
Events reach the control screen and the person on duty, and end up in statistics and reportsFeatures
Vehicles and road hazards, read live from the video
Live traffic counting
Vehicles are detected and classified from the footage, counted by type and turned into statistics.
- Counting by type
- Passing vehicles sorted by type and counted automatically
- By direction at junctions
- Counts kept separately for each direction through the junction
- Fixed or mobile
- Works for permanently installed and for mobile surveys
Live pothole detection
Potholes are found live in the driving footage, and their location and details go to the control system.
- Whatever shape they take
- Trained on potholes of many shapes and sizes
- Detected and sent live
- Found while driving, and reported as it happens
- Control room integration
- A system to monitor and manage everything that has been found
TMS · Traffic Measurement Solution
Traffic counted by vehicle type, automatically, as statistics
Vision AI tells the vehicles apart and counts them by direction of travel.
Training data
Trained across many road environments to detect more reliably
Vehicle classes
Thirteen classes in all — the ministry's twelve, plus buses
Direction tracking
Counts tracked according to the direction taken through the junction
Traffic survey system
Fixed and mobile cameras counting by vehicle type
Signal control
The counts feed the traffic signal control system
Parking management
Bay occupancy and vehicles entering and leaving, identified to support parking management
Incident Detection
Hazards on camera, turned into events for the control room
AI incident detection classifies what it sees and alerts both the system and the person on duty.
Wrong-way drivingStopped vehicleCongestionPedestrian on the carriagewayFallen objectPoor visibility or smokeRoad surface damage (pothole)Volume and vehicle type
Designed against false alarms
Confidence score, consecutive frames and human review before it is confirmed
Night and bad weather
Detection in low light, backlight and rain, corrected with real operating data
Protecting personal image data
Blurring, separated access rights and retention limits keep it within the rules
Ideal Use Cases
It suits places like these
For places that need more out of the cameras already on the poles.
Control rooms where someone must keep watching
Too many screens to follow live, so footage only gets checked after the event
Where traffic surveys are done by hand
Roads and junctions where the people and the weeks it takes are the problem
Where road inspection depends on patrols
Potholes and cracks that ought to be found before anyone reports them
Where image data protection matters
The original footage never leaves, and personal data is masked where it is filmed