Computer vision · Licence plates
The barrier reads the plate,not the driver.
Licence plate recognition (ANPR or LPR) is the computer vision technique that finds a vehicle's number plate in an image and reads its characters as text. We build systems that read plates at gates, barriers and car parks, check them against your lists and pass the result to your access or record-keeping applications, with rules on purpose and retention agreed from the start.
Events
- 07:42:13Vehicle detected at Gate 1
- 07:42:14Plate read · shown masked
- 07:42:14On the access list · access granted
- 07:42:21Entry saved to the log
Gate records are still kept on paper
At many gates, each vehicle's number is written down by hand or typed into an application. This slows entry, causes transcription mistakes and turns a simple question such as "when did this lorry arrive?" into a search through logbooks or hours of footage.
- The security guard writes plates down by hand, and queues build up at the barrier at busy times
- Transcription mistakes make the records hard to use for invoicing or reconciliation
- List-based access depends on the memory and attention of the person at the gate
- Entry and exit data does not reach the WMS, the ERP or the booking application automatically
Who it is for
- Warehouses and logistics hubs with gates that lorries and vans pass through every day
- Managers of private, office or residential car parks with list-based access
- Manufacturing sites that keep a record of delivery vehicles and visitors arriving by car
- IT teams that need a plate-reading component in their own application
Step by step
How the system reads a plate, step by step
The animation follows a single passage through the gate, from the raw camera image to the person's decision. Each of the five steps below matches one stage of the analysis.
Events
Events
- 07:42:13Vehicle detected at Gate 1
Events
- 07:42:13Vehicle detected at Gate 1
- 07:42:14Plate read · shown masked
Events
- 07:42:13Vehicle detected at Gate 1
- 07:42:14Plate read · shown masked
- 07:42:14On the access list · access granted
- 07:42:21Entry saved to the log
Events
- 07:42:13Vehicle detected at Gate 1
- 07:42:14Plate read · shown masked
- 07:42:14On the access list · access granted
- 07:42:21Entry saved to the log
Step 1 of 5
The image at the gate
The camera is mounted so that the plate appears as straight as possible, large enough and not dazzled by headlights. A usable frame has sharp characters with no motion blur; at night, infrared illumination makes the reflective plate stand out. Whatever cannot be seen clearly in the frame cannot be read later on.
Technical options and when they fit
The camera frames the plate
Suits a barrier where the vehicle always stops in roughly the same place. The camera is close with a narrow field of view, the plate fills a large part of the frame, and the system goes straight to finding and reading it.
The camera sees the vehicle
Suits wide gates, several lanes or entrances where vehicles do not stop. The system finds the vehicle first and then the plate on it, which also allows the direction of travel to be worked out and passages to be counted.
Next to the camera or on a server
The models can run on a small device by the gate, so the barrier does not depend on the internet, or on a local server that handles several gates. We export them in a format optimised for the chosen hardware.
What you receive at hand-over
- The plate-finding and plate-reading models, tuned on images from your gate and exported for the chosen hardware
- The cameras, the reading zone and the access rules configured for each vehicle list
- Integration with the barrier or the access control system and, typically, with the booking application or the ERP
- The interface for the security officer, with the cases to confirm and a passage log searchable by plate
- Retention, automatic deletion and role-based access rules, configured according to the decisions made with your data protection officer
- A measurement report on real passages at your gate, documentation and training for your team
No strings attached: tell us what the cameras should see and we'll tell you honestly whether it can be done.
Let's talkWhat the plate-reading system can do
- 01
Finds the plate in the image
The model locates the number plate on the vehicle, even when it does not stop in exactly the same spot.
- 02
Reads the characters as text
Letters and digits are turned into text, with a confidence score for each reading.
- 03
Checks against your lists
The plate read is checked against a list of authorised, booked or flagged vehicles that you define.
- 04
Triggers the agreed action
It can send the barrier an open command or a notification to the gate, following the agreed rules.
- 05
Keeps an entry and exit log
Each passage is recorded with time, camera and direction, so it can be searched by plate later.
- 06
Deletes data on schedule
Records are kept only for the agreed period and deleted automatically, and access to them is limited by role.
How it works
From a vehicle at the gate to an access decision
The steps marked in yellow are approved by a person.
- Input
- AI
- Action
- Human in the loop
- Result
- 01
Input
Vehicle at the gate
A camera aimed at the stopping area captures the front or rear of the vehicle.
- 02
AI
Locate and read
The model finds the plate and reads the characters, with a confidence score.
- 03
Action
List check
The plate is compared with the list of authorised or booked vehicles.
- 04
Human in the loop
Uncertain cases confirmed
Uncertain readings or unknown plates go to the gate officer, who decides.
- 05
Action
Command and logging
The barrier receives the command and the passage is saved in the log.
- 06
Result
Searchable record
You have a history of entries and exits, kept only as long as needed.
What plate reading needs
- Cameras mounted at a suitable angle and distance for reading the plate, ideally with infrared illumination for night-time
- Real images from your gate, by day and by night, in rain or against the light, for testing and tuning
- Vehicle lists and access rules: who enters automatically, who is confirmed by security, what happens with an unknown plate
- Access to the barrier controller or the access control system, where automatic opening is wanted
- Written decisions on purpose, retention period and who can access the log, made together with your data protection officer
Possible integrations
- IP cameras with an RTSP stream
- ONVIF-compatible cameras
- Barrier controllers and gate automation via relays or API
- Access control systems
- NVRs and VMS systems
- NVIDIA Jetson edge devices
- WMS and dock booking applications
- REST API and webhooks
- PostgreSQL databases
- IP cameras with an RTSP stream
- ONVIF-compatible cameras
- Barrier controllers and gate automation via relays or API
- Access control systems
- NVRs and VMS systems
- NVIDIA Jetson edge devices
- WMS and dock booking applications
- REST API and webhooks
- PostgreSQL databases
Each integration's compatibility is checked during the analysis phase, for the systems and versions you use.
Control and transparency
What stays with people
- Setting the purpose, the retention period and how drivers are informed, for example with signs at the entrance
- Access decisions for uncertain readings, unknown plates or vehicles without a visible plate
- Updating the lists of authorised vehicles and regularly reviewing who has access to the log
- Approving any use of the data for a purpose other than the one originally set
Limits and operating conditions
- Dirty, damaged or snow-covered plates, or plates partly hidden by trailers and tow bars, are harder or impossible to read
- Steep angles, vehicle speed and headlights pointing at the camera can reduce reading quality, especially at night
- Plates from other countries, temporary plates or special formats need separate testing and sometimes extra examples
- Licence plates are personal data; the system must be used only for the stated purpose, with people informed and limited retention periods
How we evaluate plate-reading quality
We agree the indicators together before launch and track them afterwards. The values depend on the process, the data and the volume.
- Fully correct readings
- The share of plates read correctly character by character, checked manually on a sample of real passages at your gate
- Vehicles with no reading
- Passages in which the plate was not found or not read, counted on the same sample and grouped by cause
- Character confusions
- Pairs of characters that are often confused, such as 0 and O or 8 and B, analysed on the test set
- Stability across conditions
- Results compared separately for day and night, good weather and rain, cars and lorries
How we put plate reading into service, in 4 steps
- 01
We set the purpose and the rules
We define what problem the system solves, which data is kept, for how long and who sees it, before anything is installed.
- 02
We check the gate and the cameras
We review the angle, distance, lighting and stream of the existing cameras, or propose the right position for a new one.
- 03
We configure and test on real traffic
We run reading on real passages, measure the errors and tune until the results are stable.
- 04
We integrate and monitor
We connect the result to the barrier and your applications, then monitor reading quality over time.
Use cases
The scenarios below are conceptual examples, designed to illustrate realistic applications. They are not implementations delivered by WEB-AZ.
- Logistics
Warehouse gate linked to bookings
A lorry booked for a dock is recognised at the gate, and the dispatcher sees the arrival in the booking application; unbooked vehicles wait for security to confirm.
Conceptual example - Office buildings
Car park for staff and visitors
Registered staff plates open the barrier, reception is notified about visitors, and the log is deleted automatically after the agreed period.
Conceptual example - Manufacturing
Recording delivery vehicles
The entry and exit time of each delivery vehicle goes into the ERP, where the purchasing team uses it to reconcile goods received.
Conceptual example
Useful guides
Before you decide, read
Frequently asked questions
What is licence plate recognition (ANPR)?
It is a computer vision technique in which a system finds a vehicle's number plate in an image and reads its characters as text. The result can be checked against a list, used to control a barrier or saved in an entry and exit log.
Are licence plates personal data?
Yes, they are generally treated as personal data because they can be linked to a person. That is why we agree the purpose, legal basis, retention period and how drivers are informed from the start, together with your data protection officer. For your specific situation, we also recommend a legal review.
Do you identify the driver or passengers?
No. The system reads only the vehicle's plate and does not use facial recognition or any other form of biometric identification. The camera is aimed at the plate area, and where people appear in the image, stored frames can be cropped or blurred.
Does it work at night and in bad weather?
It can, but quality depends on the camera and the lighting. Infrared cameras help a lot at night, while rain, snow or strong headlights can reduce reading quality. We measure results separately for these conditions before going live.
Can I use my existing CCTV cameras?
Sometimes, but a camera installed for general surveillance often lacks the right angle or resolution for plates. We check the actual images from your gate and tell you whether they are enough or which camera would suit.
What happens when a plate is not read correctly?
Each reading has a confidence score. When it is low or the plate is not on the list, the decision goes to the person at the gate, who can enter or correct the plate. The barrier does not open automatically on an uncertain reading.
Can’t find the answer? Write to us and we’ll reply directly.
Do you have a gate where vehicle records are kept by hand?
Send us a few images from the gate, by day and by night, and tell us what should happen at each passage. We will tell you what can be automated and which data rules should be set.
Related solutions
- Car park monitoringOccupancy of every parking space or zone, read from cameras, with free-space counts kept up to date and no identification of drivers.
- Vehicle traffic analysisAnonymous counting and classification of vehicles by direction, at gates, on access roads or on building sites, with data aggregated by interval.
- Container code recognitionAutomatic reading of ISO 6346 codes on containers, with check digit validation and direct transfer to the terminal system.