For control rooms and investigators working across cameras they do not all own. Search by plate or description, review the candidate sightings, and check each one against its still frame and your recorder before it goes into a file.
- 1Northern approach · 10:41 · first seen
- 2Market square · 10:43 · seen again
- 3Depot gate · 10:44 · seen at the gate
- 4Southern approach · 10:46 · last seen
The problems we start from
Who it is for
- Control-room operators
- Investigating officers
- Traffic branch
- District and range leadership
Tracing a vehicle means calling station after station
Footage sits in separate recorders. Finding one vehicle across them takes days of requests and manual review.
Alerts arrive without context
An alert with no picture, no history and no owner is dismissed or handled twice.
No record of who searched for what
A sensitive search needs an owner, a purpose and a trail that a supervisor can read.
One workflow, step by step
Trace a reported vehicle
Scenario. A vehicle is reported leaving the scene of an incident. An investigator has a partial plate and a time.
Set the scope
Choose the time window. The search covers only the cameras your organisation owns or has been granted, narrowed by your own camera scope.
Search by plate or description
Enter the plate, or the part of it you have, or filter by vehicle type and colour.
Review the candidates
Each sighting shows its still frame, camera, time, the plate as read and whether the match was exact or near. A near match is a candidate, not a confirmation.
See the route
Readable sightings appear as numbered stops in time order on a map. Lines between stops are inferred; a gap is where no camera read the plate.
Check the source
Open the still for each stop. For the moments around it, go to the recorder that holds that camera's video.
Record the finding
Today a finding is recorded in the product only as a note on an alert. A trace that starts from a search has no alert, so the finding goes in your own case file; the plate search itself is written to the audit log.
Measures
Four measures, each with a definition.
Defined so that two people would compute the same number. None is a result: the values come from your pilot.
- Time to first useful evidence
- From the start of a search to the first sighting the investigator notes as relevant.
- Measured from: Timed by hand in the pilot; the audit log gives the time of a plate, scene or route search.
- Plate read rate
- Sightings with a readable plate divided by all vehicle sightings, per camera, split by day and night.
- Measured from: Plate statistics in Insights.
- Pending reviews
- Alerts still marked New after a time you agree.
- Measured from: The alert queue, by status.
- False-alarm rate
- Alerts marked False positive divided by alerts reviewed, per rule and per camera.
- Measured from: Alert statistics in Insights.
Events of interest
Events a control room can switch on.
Each says what it needs and where it stops. The rules you set on a camera raise these events, and the alerts that go with them.
Plate on watchlist
A plate that is read matches an entry on your organisation's watchlist.
- When
- A plate read matches a watched plate exactly, with commonly confused characters treated as the same, or within one character.
- Limit
- A plate that cannot be read cannot be matched. A near match must be compared by a person before anyone acts.
Wrong-way movement
A vehicle, or a person on foot, crosses a line against the permitted direction. Each rule names the object type it applies to.
- When
- A line crossed in the direction the rule forbids.
- Limit
- A vehicle reversing briefly, or a person stepping back, crosses the line the wrong way.
Rider without helmet
A scene reading describes a two-wheeler rider or passenger with no helmet.
- When
- The reading of a two-wheeler sighting carries the no-helmet tag.
- Limit
- Caps, hoods and distant riders are read wrongly at times. Treat it as a candidate for review, not a finding.
Triple riding
A scene reading describes three or more people on one two-wheeler.
- When
- The reading of a two-wheeler sighting carries the triple-riding tag.
- Limit
- Two riders close to a pedestrian, or two motorcycles side by side, can be read as three.
Vehicle stopped where it should not
A vehicle stays in a no-stopping zone, a fire lane or an emergency bay beyond the time you set.
- When
- A vehicle's time inside the zone passes the limit.
- Limit
- A vehicle held in traffic inside the zone is reported like a parked one.
Crowd gathering
More people are in a camera's view at once than the number you set.
- When
- The count of people in the camera's view reaches the threshold.
- Limit
- A count of people the camera can see, not a calibrated density and not a count inside a drawn zone. People hidden behind others are not counted.
Vehicle counts by type
Cars, buses, trucks, motorcycles and bicycles counted for each camera, by quarter-hour, hour or day.
- When
- Each followed vehicle is counted once in every period in which it is seen.
- Limit
- A vehicle that leaves the view and returns is counted again, and one seen across two periods is counted in each. Vehicle types outside the detected set are counted as the nearest type or not at all. Counts cover the last 30 days.
Loitering
A person stays in an area longer than the time you set.
- When
- Time inside the zone passes the limit.
- Limit
- If a person is lost and found again, the timer can restart, so long stays may be reported late.
Camera offline
A feed stops arriving. The platform keeps trying to reconnect and shows the camera as offline until the picture returns.
- When
- The feed is still missing after repeated reconnection attempts.
- Limit
- A camera that streams a frozen picture can look healthy. The alert repeats while the feed stays down.
Use cases
What investigators and operators use it for.
What the current release does for this kind of site.
Trace a vehicle by plate
Readable sightings in time order, as stops on a map
Trace back by appearance
Look-alike sightings grouped as candidates across cameras
Road-safety candidates
Wrong-way movement, riders without helmets and triple riding brought to a reviewer
Crowd gathering
An alert when the people in view pass a threshold
Search uploaded footage
Exported clips analysed and searched at the time they were recorded
Share cameras between agencies
Each camera shared with named organisations, each with its own alerts
Weekly situation brief
A fact sheet and written summary for each organisation
Watch for a plate
An alert when a watched plate is read
What it needs
- Stream addresses for the cameras in scope, from each agency that owns them
- Camera locations entered for every camera that should appear on the map
- Cameras placed so that plates are legible, with the read rate measured before it is relied on
- Named roles for search, triage and administration
Pilot objective
On a corridor or a set of junctions you choose, show whether a reported vehicle can be traced across cameras from its plate or description, and measure the plate read rate and the false-alarm rate on that footage.
Read the pilot guideBoundaries
- A match is a candidate. VisionObject does not identify a person or conclude that an offence took place.
- Access is bound to organisation, role and camera scope, and searches by plate, scene description and route are written to the audit log.
- There is no case file, evidence export or handling record in the product, and stills are not cryptographically sealed.
- Vehicle-registry lookups, dispatch and penalty-notice connections are not packaged. Each is scoped as a separate integration.
- Device location is not used. Drone and body-worn footage would be handled as uploaded files only; it has not been tested and requires assessment.
- Cameras are shared between organisations on one deployment. Separate deployments are not joined.
Capabilities used
The same platform. Different rules.
Each site gets its own cameras, modules and rules. Nothing is custom-built; everything is configured.
Object and vehicle recognition
Detect people, vehicles and carried bags, read plates where they are legible, and find similar-looking sightings.
Object classification
Record what can be observed about each sighting: its type, colours, tags, plate and where and when it was seen.
Advanced video analytics
Analytics that go beyond a single camera or a single frame, each with what it measures and how far it has been evaluated.
Trace one vehicle across your own cameras.
Choose a corridor or a set of junctions and a past incident. We will show what can be read, linked and reviewed on that footage, and what cannot.