Computer vision · Car park monitoring
Free spaces,seen at a glance.
Car park monitoring with computer vision uses cameras and AI to find out, for each space or each zone, whether it is occupied or free. We build systems that count free spaces and measure occupancy and length of stay, without reading licence plates where that is not necessary.
Events
- 16:44:02Zone B · 10 of 16 spaces taken
- 16:44:026 free spaces shown at the entrance
- 16:44:05Vehicle searching in the lane
- 16:44:05No plate reading in this zone
Nobody knows exactly how many spaces are free
Without continuous measurement, drivers look for a space at random, and the operator learns about occupancy from estimates or complaints. Sensors fitted to each space are one solution, but they need installation work and maintenance on every space, while the existing cameras are used only for recording.
- Drivers circle the car park looking for a space, and entrances get congested at peak times
- Reserved spaces, disabled bays or EV charging bays are occupied without anyone noticing
- Decisions on expansion, tariffs or rules are made without data on how the car park is used by hour and day
- Surveillance cameras film the car park but pass on no information about occupancy
Who it is for
- Operators of car parks, shopping centres, office buildings or hospitals that want to display free spaces
- Companies with car parks for staff, fleets or visitors that want to know how much the space is used
- Logistics centres with waiting areas for lorries and a limited number of loading bays
- Local authorities and urban planning firms that need aggregated data on car park occupancy
Step by step
How the system finds free spaces, step by step
The five stages on the animation follow a single camera above one zone of a car park: from the raw image to the number of free spaces shown at the entrance.
Events
Events
- 16:44:02Zone B · 10 of 16 spaces taken
Events
- 16:44:02Zone B · 10 of 16 spaces taken
- 16:44:026 free spaces shown at the entrance
Events
- 16:44:02Zone B · 10 of 16 spaces taken
- 16:44:026 free spaces shown at the entrance
- 16:44:05Vehicle searching in the lane
- 16:44:05No plate reading in this zone
Events
- 16:44:02Zone B · 10 of 16 spaces taken
- 16:44:026 free spaces shown at the entrance
- 16:44:05Vehicle searching in the lane
- 16:44:05No plate reading in this zone
Step 1 of 5
The camera above the spaces
A camera mounted high up, on a pole or a building, sees the spaces from above or from a wide angle, so that one vehicle does not hide the ones behind it. We draw each space and each zone on the image once, together with their capacity. At night, the car park lighting or the camera's infrared decides how clearly each space can be seen.
Three ways to measure occupancy
Space by space
Where the markings are clearly visible and the camera has a good view of them, each space is drawn separately and gets its own status. This is the right option for guiding drivers row by row and for spaces with special rules.
By zone, against capacity
Where markings are missing, covered by snow or the camera sees the car park from too far away, the system counts the vehicles in a zone and subtracts them from its capacity. The result is less detailed, but more robust for large, unmarked car parks or ones seen from a low angle.
At entrances and exits
For car parks with few access points, a camera at each entrance can count the vehicles crossing a virtual line in each direction to work out how many are inside. It needs fewer cameras, but it does not show where the free space actually is, and the total is recalibrated periodically.
What you receive at hand-over
- The vehicle detection model adapted to your car park, exported for the edge device or local server
- The spaces, zones, capacities and rules for special spaces, drawn and configured on each camera
- Integration with the entrance signs, web map or your app, over an API or MQTT
- A staff dashboard with the flags to check and occupancy statistics by hour, day and zone
- A measurement report from the test period, comparing space status with reality by day, by night and in bad weather
- Documentation, training and, typically, a recalibration procedure for when a camera is moved or the markings are repainted
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 car park monitoring system can do
- 01
Occupancy of every space
Parking spaces are drawn on the image, and the system shows for each one whether it is free or occupied by a vehicle.
- 02
Free spaces by zone
Where markings are missing or not visible, the system estimates how many vehicles are in a zone and how many spaces are left, based on the agreed capacity.
- 03
Length of stay
Measures how long a space stays occupied, without identifying the vehicle, to see how spaces turn over and spot overstays.
- 04
Flags for special spaces
Can flag when a charging bay, a delivery bay or an access route is occupied for longer than allowed, for staff to check.
- 05
Data for displays and apps
The number of free spaces goes via API to the signs at the entrance, a web map or your application.
- 06
Occupancy statistics
Occupancy is aggregated by hour, day and zone, so you know when the car park is full and when it stays empty.
How it works
From the camera image to the free spaces on display
The steps marked in yellow are approved by a person.
- Input
- AI
- Action
- Human in the loop
- Result
- 01
Input
Car park images
The cameras covering the spaces or zones send frames at set intervals.
- 02
AI
Vehicle detection
The model finds vehicles and maps them to the spaces or zones drawn.
- 03
Action
Occupancy calculation
The status of each space and the free-space count by zone are updated.
- 04
Human in the loop
Exception review
Staff check flags, for example a special space occupied for too long.
- 05
Result
Free spaces and reports
Signs, a map or an app with free spaces and occupancy statistics.
What car park monitoring needs
- Cameras positioned to see the spaces or zones of interest, ideally from height, with as few obstacles as possible between them and the vehicles
- The car park plan, with spaces, zones, the capacity of each and the spaces with special rules
- Real images from the car park, at different times and in different conditions: day, night, rain, snow
- Access to camera streams (for example via RTSP or ONVIF, where the equipment supports them)
- An edge device or local server for processing, with a link to the displays or applications
Possible integrations
- IP cameras with an RTSP stream
- ONVIF-compatible cameras
- NVRs and VMS systems
- NVIDIA Jetson edge devices
- Free-space display signs, via API or a dedicated controller
- Parking management systems and barriers, where they offer an integration interface
- Web maps and mobile apps, via REST API
- MQTT
- PostgreSQL databases
- Reports in Power BI or Excel
- IP cameras with an RTSP stream
- ONVIF-compatible cameras
- NVRs and VMS systems
- NVIDIA Jetson edge devices
- Free-space display signs, via API or a dedicated controller
- Parking management systems and barriers, where they offer an integration interface
- Web maps and mobile apps, via REST API
- MQTT
- PostgreSQL databases
- Reports in Power BI or Excel
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 rules: what counts as an overstay, which spaces have special rules and who receives the flags
- Checking any flag on site before taking action towards a driver
- Deciding to add licence plate reading, only where there is a clear purpose and a legal basis
- Periodically checking the calibration of spaces on the image, after a camera is moved or the markings are repainted
Limits and operating conditions
- Tall vehicles, posts or trees can hide the spaces behind them; coverage depends on the position and height of the cameras
- Heavy rain, fog, snow or poor lighting at night can reduce detection quality
- Without plate reading, the system does not know which vehicle occupied a space and cannot follow the same vehicle between cameras
- One camera can only cover a limited number of spaces with a clear enough image; large car parks need several cameras
How we evaluate car park monitoring accuracy
We agree the indicators together before launch and track them afterwards. The values depend on the process, the data and the volume.
- Correct space status
- The status the system shows for each space, compared with a manual check on images or on site, at different times
- Error in the free-space count
- The difference between the number displayed and the real number by zone, measured on a sample of checked moments
- Update time
- The interval between a space becoming free and the display changing, measured in operation
- Stability across conditions
- Results compared separately for day, night, rain and busy periods
How we implement car park monitoring, in 4 steps
- 01
We review the car park and cameras
We check what the existing cameras see, where others would be needed and what information needs to reach drivers or the operator.
- 02
We draw the spaces and rules
We mark the spaces, zones and their capacity on the image and set the rules for special spaces and overstays.
- 03
We test in parallel
The system runs for a period alongside the usual checks, and the results are compared with reality in different conditions.
- 04
We connect the displays and monitor
We link the data to signs, apps or reports and monitor quality over time, adjusting when the car park changes.
Use cases
The scenarios below are conceptual examples, designed to illustrate realistic applications. They are not implementations delivered by WEB-AZ.
- Retail
Free spaces at a shopping centre entrance
Pole-mounted cameras cover the car park zones, and the sign at the entrance shows how many free spaces there are in each row, without vehicle plates being read.
Conceptual example - Healthcare
A hospital car park
The management sees occupancy by hour and day, and staff receive a flag when the ambulance access route or a disabled bay is occupied for a long time.
Conceptual example - Logistics
The lorry waiting area
The dispatcher sees how many waiting spaces are occupied and for how long, to plan when lorries are called to the loading bays.
Conceptual example
Useful guides
Before you decide, read
Frequently asked questions
What is car park monitoring with cameras and AI?
It is the use of cameras and a computer vision model to find out, continuously, which parking spaces are occupied and which are free. The result is the number of free spaces by zone, occupancy and length of stay, sent to displays, apps or reports.
Does the system read licence plates?
No, in its basic form it only detects whether vehicles are present in spaces or zones. Licence plates are personal data, so reading them is added only where there is a clear purpose, for example access to a permit-holder car park, a legal basis and rules on retention and notice.
Can I use my existing surveillance cameras?
Sometimes, if they see the spaces from a suitable height and angle and provide an accessible stream. We check each camera in practice, and where coverage falls short we recommend positions or additional cameras.
How does it differ from sensors fitted to each space?
Sensors in each space give direct information for that space, but need installation and maintenance on every single one. A camera can cover several spaces at once and can use existing infrastructure, but it depends on visibility and weather. The choice depends on the car park, and we discuss it openly.
Does it work at night or in the rain?
It can, but quality depends on the lighting and the camera. That is why we also test the system on images taken at night, in rain or snow, and measure the results separately for these conditions before putting it into operation.
What happens to the car park images?
The analysis can run locally, and only the status of spaces and aggregated numbers are passed on. Where images need to be kept, the period and access are agreed in advance, and people who appear in the frame can be blurred. A filmed car park remains subject to the usual obligations on video surveillance and data protection.
Can’t find the answer? Write to us and we’ll reply directly.
Want to know at every moment how many spaces are free?
Send us the car park plan and a few images from the existing cameras. We will tell you what they can cover and what else would be needed.
Related solutions
- Licence plate recognitionAutomatic reading of licence plates at gates, barriers and car parks, with clear rules on access, retention and human review.
- Vehicle traffic analysisAnonymous counting and classification of vehicles by direction, at gates, on access roads or on building sites, with data aggregated by interval.
- Edge AI for computer visionVideo analysis on devices installed on site: fast response, less network load and images that never leave the building.