Vehicle make and model
Recognizes the make, model and body year of cars; stores the three most likely results together.
Vehicle make and model analytics examines the cars in the image and identifies each vehicle’s make, model and body year. ProVife stores the three most likely results for each vehicle; if the top two results are very close, no event is produced for the uncertain vehicle. An event is created when the result repeats for the consecutive confirmation count. It is used to detect vehicles with swapped or fake plates, to compare the vehicle information in the plate record with the vehicle in the image, and to search recordings for a specific model. The analytic works on cars.
How it works
Draw the area where cars will be seen.
Enter the confidence threshold and confirmation count for the make and model decision.
ProVife examines every car that appears large enough and extracts the three most likely make, model and year results.
If the top two results are very close, no event is produced for the vehicle.
When the same make and model result repeats enough times, the event is recorded.
Settings
| Area | The zone where cars are examined. |
|---|---|
| Confidence threshold | The minimum confidence required to accept a make and model result. |
| Consecutive confirmation | How many evaluations must return the same make and model result. |
| Repeat interval | The minimum time between two events for the same car. |
What it produces
- Vehicle make and model event (analytics.vehicle.make_model): make, model, year and confidence value.
- The three most likely results and the confidence value of each.
- A make and model label over the vehicle in live video.
- Event image with the car marked by a red box and the model name.
What you can do with the event
- Finding related vehicle passes by typing a make and model in the recordings.
- Comparing plate and vehicle information by reviewing alongside plate reading events.
- Forwarding events to external systems via MQTT.
- Sharing the make and model breakdown in a PDF report.
Where it is used
Public sector
A wanted vehicle model is quickly found in recordings by date and camera.
Car dealerships
The make and model breakdown of vehicles coming in for service is tracked.
Parking lot
A subscriber vehicle’s plate is reviewed together with the model of the vehicle in the image.
Residential complexes
Fake plates put on residents’ vehicles are caught by the vehicle model.
Camera placement
- Position the camera to see the car from the front or front diagonal, including the front bumper and grille.
- Choose a close distance where the vehicle fills a large part of the image.
- Spots in front of barriers where vehicles stop give the clearest results.
- Block backlight so the front of the vehicle stays well lit.