- Home
- Resources
- License plate recognition for warehouses and plants: what it gives, which cameras work, what drives accuracy
License plate recognition for warehouses and plants: what it gives, which cameras work, what drives accuracy
Author: DuoVision team
For: heads of warehouses, plants, logistics and security who want to log vehicles at the gate by plate using cameras they already have
Summary
- Plate recognition records the plate, date, time, camera and a photo for every vehicle. Those rows make an entry and exit log, time on site and at the ramp, and white and black lists.
- A camera that is already at the gate works if the plate in the frame is large enough, sharp and not at a steep angle. The Dahua guide asks for at least 130 pixels of plate width for a single-row plate (it does not name Ukraine, so we use its 'all other countries' row) and an angle to the direction of travel up to 30 degrees. This is checked on live video from your site.
- Accuracy depends on dirt on the plate, night light, camera angle, vehicle speed and whether a person handles the disputed passes. There is no universal accuracy percentage, so in a pilot we show the result on your cameras before launch.
- A plate together with time and place can count as personal data (indirect identification, according to the Ukrainian Ombudsman), so before launch you decide the purpose, who has access and how long the log is kept.
For: heads of warehouses, plants, logistics and security who want to log vehicles at the gate by plate using cameras they already have
What it is and what it gives a business
A camera at the gate films the vehicle, software reads the plate from the frame, and each pass lands in a log as a row: plate, date and time, camera, direction and photo. That is all recognition does by itself. The value comes from the reports a business builds on those rows.
- An entry and exit log with photos. A disputed pass is opened by plate or date instead of scrubbing through video.
- Time on site. The gap between the entry row and the exit row shows how long a vehicle stayed.
- Time at the ramp, if a camera also covers the ramp doors. One plate across several rows splits a truck's visit into stages: gate, scales, ramp, exit.
- White and black lists. Your own and approved vehicles pass as usual, and an unknown plate outside working hours goes into a separate card.
- Matching against delivery notes or passes. It works if the list of passes or documents can be taken from your accounting or weighbridge software (export, database, API). We check that on site before launch. Without such a list the system keeps the log, which is enough to start.
The guard stays. He stops copying plates by hand and handles the cases the system passed to a person. Lists, cards in the panel and the ramp queue are covered in the article on tracking vehicles on a plant site, and how such a system is built in license plate recognition system.
Which cameras work
The plate is read by a camera aimed at the lane the vehicle drives along. An overview camera above the yard sees everything, but the plate in its frame is usually small. Manufacturers publish their own requirements, and we rely on the Dahua guide "LPR Camera Selection & Installation Guide v1.1", which we checked on 12.11.2026. For other brands the numbers can differ, so they are checked against the documentation of the specific model.
- Plate size in the frame. For a single-row plate the guide requires at least 130 pixels of width (the "all other countries" row; Ukraine is not named separately). For a double-row plate the minimum is 70. The guide adds that most failures come from a plate that is too small, and a slightly larger one does no harm.
- Resolution and lanes. According to the guide, for European and CIS plates 1080P covers 2 lanes and 4 MP covers 3 lanes.
- Angle. The horizontal angle between the camera and the direction of travel is up to 30 degrees, and the vertical angle is also up to 30. Tilt of the plate itself in the frame should stay within plus or minus 5 degrees.
- Height and distance. For monitoring at 4 to 8 meters the guide gives a camera height of 1 to 2 meters. For 10 to 40 meters the camera goes up to 4 to 6 meters, but the plate then gets smaller.
- Vehicle speed. At a barrier where the vehicle stops, motion blurs the frame less. If vehicles drive fast, the guide advises shortening the shutter to about 2.5 to 3 ms.
- Night. Infrared illumination is needed. If the plate is overexposed, the guide advises reducing the IR lamp power or the shutter.
In practice we take video from your camera by day and by night and measure how many pixels the plate occupies at the distance you need. If it is enough, no new camera is needed. If not, one plate camera is added per lane and the overview cameras stay where they are.
What accuracy depends on
No system reads every plate, and the honest answer to a question about accuracy is that it depends on your gate. What changes it:
- Condition of the plate. Dirt, snow, a bent plate, a trailer or cargo covering it.
- Light. At night the plate is read in infrared, and too strong illumination washes it out.
- Geometry. A camera angle that is too steep, a plate that is too small in the frame, a tilted plate.
- Speed. Where vehicles pass without stopping, motion blurs the frame.
- Plate type. Regular Ukrainian plates, transit, temporary, military and foreign plates look different, and each type is checked on your own frames.
So the system has to admit when it is unsure. An unreadable pass stays in the log marked "not read" with its frame, and a person types the plate in. The share of such passes is visible only on your traffic, and that is what we show in a pilot, together with the number of passes processed.
Ukrainian plates use Cyrillic letters that look the same as Latin ones (А, В, Е, І, К, М, Н, О, Р, С, Т, Х), and the format is two letters of the region code, four digits, two letters of the series. Keep one form of the record in the log and normalize letters when searching, otherwise Cyrillic А and Latin A give two different plates. Sources: MIA Order No. 166 of 02.03.2021 and Wikipedia (the page is flagged as needing an update).
Personal data: what to do with the plate and the time
In its guidance on CCTV, the Ukrainian Parliament Commissioner for Human Rights says a person can be identified directly (recognizing someone in the car) or indirectly, through the state license plate. The Law "On Personal Data Protection" (article 2) defines personal data as information about a person who is identified or can be specifically identified. So a log that holds a plate, a time and a place is best treated as personal data. That is settled with four decisions written down once.
- Purpose. For example vehicle tracking on site, security, access control. The data must fit the purpose and not be excessive (article 6).
- Access. Who sees the log and photos and who may export them.
- Retention. The law gives no fixed number of days. Data is deleted when the period set by consent or by law ends (article 15), In practice the business sets the period in its own policy based on the purpose. The Commissioner advises stating the retention period and the deletion procedure in the processing documents.
- Notice. The Commissioner names a sign, a leaflet or a website as ways to inform people, so a video surveillance sign usually goes up at the gate, with the processing terms published in a policy.
We do not recognize faces, and a plate describes only the vehicle. More on the law and employer requirements in video analytics and personal data.
What drives the price
There is no single price for plate recognition because several parts add up. We give no sums here. Below is what forms the price, so you can compare offers on the same basis.
- Number of cameras and lanes. According to the Dahua guide, one camera reads 2 to 3 lanes depending on resolution.
- Whether existing cameras fit. If they do, no cameras need buying. If not, a camera with illumination, a bracket and installation are added.
- Payment model: a monthly subscription per camera or a one-off license, and what the sum covers (archive, number of cameras, vehicle make and color).
- Where video is processed: on a server at your site, in your cloud or in someone else's. That drives hardware and network requirements.
- Integrations. Connection to the weighbridge, accounting software, a pass system or a barrier.
- Log and photo retention, because it sets the storage volume.
Market tariff examples and what they include are in the article on plates at warehouse and grain elevator gates.
How we launch
We recognize vehicle plates on the cameras already at your site, and we add a camera only where video from the existing one did not give a readable plate.
- We pick one gate: the busiest or the most disputed.
- We take video from your camera by day and by night, measure the plate size in the frame and check whether it reads.
- We connect to the existing camera. If the plate is too small or the angle too steep, we propose one camera per lane.
- We show the result on your cameras before launch: what was recognized, what went to "not read" and why.
- We run in shadow mode: only you see the log, with no alerts to staff. We went through that order at a production site where we launched the system earlier: there we counted half carcasses and bags, and we did not recognize vehicle plates there.
- Then we decide whether to connect lists, matching with documents, the exit and further gates.
Video is processed on a server at your site when your security policy requires it. To ask about a pilot, use the contacts page. For industry use see ramps and logistics and site security.
FAQ
Can plates be recognized on cameras that are already installed?
Sometimes. What decides it is how many pixels the plate occupies at the needed distance and the angle at which the camera looks at the lane. For a single-row plate the Dahua guide asks for 130 pixels of width (its "all other countries" row) and an angle up to 30 degrees. This is checked on video from your site by day and by night. If the plate does not read, one camera is added on the lane.
Does the system recognize plates at night, in dirt and in snow?
At night infrared illumination is needed, and too strong illumination can wash out the plate. A dirty, bent plate or one covered by a trailer may not be read. Such a pass stays in the log marked "not read" with its frame, and a person types the plate in. The share of such passes is visible on your traffic, and we show it in a pilot.
Is a vehicle plate personal data?
The Ukrainian Ombudsman treats a license plate as a way of indirect identification of a person. A log with plate, time and place is best handled as personal data: set the purpose, who has access and the retention period, and inform people, for example with a video surveillance sign. The law gives no fixed retention period, so the business sets it.
What does the price of a license plate recognition system depend on?
On the number of cameras and lanes, whether existing cameras fit, the payment model (subscription or license), where video is processed, integrations with a weighbridge or accounting software, and how long the log is kept.
Do you recognize the driver's face or the weight of the cargo?
No. We do not recognize faces. Cargo weight comes from scales, and the camera records that the vehicle drove onto them.
Where to apply this
- Warehousing and logistics
- Agriculture and grain elevators
- Meat processing
- Site security and vehicles
- Loading docks and logistics
- Case study: video hygiene control at a meat processing plant, 1,192 cards in 8 weeks
- Hard hat detection by camera: construction, warehouse and production. What it sees and where the limits are
- Work clothing control in food production: when work clothes leave the workshop
Sources
- Dahua. LPR Camera Selection & Installation Guide v1.1 (ITC237, ITC206, ITC302, ITC602), checked 12.11.2026
- Law of Ukraine "On Personal Data Protection" of 01.06.2010 No. 2297-VI (articles 2, 6, 15)
- Ukrainian Parliament Commissioner for Human Rights. Guidance "Video surveillance cameras: what everyone should know to protect their privacy" (in Ukrainian)
- Ministry of Internal Affairs of Ukraine. Order No. 166 of 02.03.2021 "On certain issues of vehicle license plates"
- Wikipedia. Vehicle registration plates of Ukraine (the page is flagged as needing an update)