For most parking platforms, the answer is the officer. Everything on either side of that field is automated, and the plate is the one place the software waits before it can go on. It is also the easiest gap to close, because closing it does not mean touching the rest of your product.
What a license plate reader adds to software you already ship
A smartphone license plate reader is software that reads plates through a phone camera. No extra hardware and no connection needed. The industry calls it smartphone ALPR, short for automatic license plate recognition.
It arrives as an SDK, which is a component your developers add to the app you already ship. Your permit database, citation rules, payment handling, reporting, and officer screens all stay exactly as they are. Your app or software already has a field for entering a plate. The reader fills it in for them.
That narrow footprint is the point. Adding plate reading is a feature release, and it does not touch the parts of your platform that took years to get right.
Why integration teams choose Anyline
Four things decide whether a plate reader survives contact with a real parking lot. Anyline was built to answer all four, and any vendor you evaluate should be made to answer them too.
1. It runs on every platform your app ships on
Anyline’s plate reader is built natively for iOS and Android. It supports React Native, Cordova, and Flutter, and there is a web SDK for JavaScript.
Ask a plate engine vendor for their iOS SDK. Many in this market cannot supply one because they are Android-only, with no cross-platform wrappers and no browser option.
If your app ships on both mobile platforms, that leaves you two poor choices: standardize every officer on Android hardware, or ship an iOS build with a feature missing.
2. Continuous scanning, so the officer keeps walking
Continuous scanning means the camera stays open and reads plates as they pass. The officer walks a row and the results land in a list behind them.
Compare that with the common pattern. Stop at the car, frame the plate, tap, wait for the answer, check it, move to the next one. Repeat that four hundred times in a shift, and you have data entry with a camera in your hand. For your developers, the continuous version is also less work, because a list that fills itself is simpler than a capture screen the user has to drive.
3. It reads on the phone, so it works with no signal
Anyline processes license plate data entirely on-device. By default, no images or results reach Anyline servers, and the SDK can run fully offline.
Level five of a concrete garage is where this gets decided. A cloud reader sends every frame to a server, so when the signal drops, it stops returning answers, and your officer is standing still. An on-device reader carries on, and your app syncs the reads when the phone finds a connection.
It shortens the procurement conversation too. When plate images never leave the phone, there is no third-party processor in the capture path for a customer’s data-protection review to work through. Operators running public contracts ask about this early.
4. One integration reads plates and VINs
The same SDK reads vehicle identification numbers. You integrate once and get both identifiers, which matters more than it sounds and has its own section below.
Anyline reads plates with over 99% accuracy. More than 100 million scans run through it a year. Coverage is US, Canadian, and EU plates with the region detected automatically, which is what an out-of-state visitor lot needs.
When there is no plate to read, the VIN identifies the vehicle
Front plates are not universal in the United States. Twenty-eight states and Washington, D.C. require both a front and a rear plate on most passenger vehicles, which means that in the rest of the country a car may legally carry a rear plate only.
Now picture that car reversed into a bay with its back to a wall. The rear plate faces the wall. No front plate was ever issued. A camera in the aisle has nothing to work with, and the car is parked perfectly legally.
The windshield, though, is now facing the aisle. That is where the VIN sits. Anyline reads it from the windshield, the door sticker or the chassis, using the same SDK you already integrated for plates. The officer identifies the vehicle and carries on down the row.
Plates go missing for duller reasons too. A dealer paper tag nobody has replaced. A plate bent in a parking knock, caked in winter salt, or hidden behind a bike rack. In each case, the VIN gives you a second route to the same answer, and no fixed camera at a gate can reach it.
This is worth weighing when you compare readers. A plate-only engine leaves you with a category of vehicles your software cannot identify, and you will not find out how large that category is until your customers are live.
How this compares to the other ways to read a plate
Five approaches are on the market. They suit different jobs, and here is the honest version.
| Approach | Best at | Where it falls short |
|---|---|---|
| Fixed cameras at entry and exit | Gates, ticketless access, timing a stay for billing | Reads one spot only. Nothing in the rows, no VIN. Cameras, mounting, power and cabling per lane |
| Camera on a patrol vehicle | Covering large lots quickly from the driver’s seat | Hardware plus a vehicle. Misses bays a car cannot drive past. Nothing on foot |
| Cloud plate-reading API | Back-office batches and server-side work | Needs a signal for every read. Stops in a garage. Plate images leave your system |
| Open-source plate library | Budget projects and clean, well-lit plates | You own the tuning, the model updates, and the awkward reads. No support path |
| Anyline smartphone SDK | Officers on foot. Every platform. Offline reads. Plates and VINs together | Built for parked vehicles, so highway speeds and gantry-scale volume belong to fixed cameras |
Within the smartphone category, four questions separate the options, and they are the four from the section above. Does it run on both mobile platforms, or only Android? Does it read on the device, or send every frame to a server? Does it capture continuously, or one photo at a time? Does it read VINs as well as plates? Put those in your evaluation, and most of the shortlist answers itself.
Plate designs change every year, so the reader has to change too
A plate reader is aiming at a moving target. States redesign plates and add specialty series constantly. One count put the total at 8,331 plate designs across the fifty states and DC in mid-2023, and its author noted that some had already changed while he was counting.
The agencies that issue the plates make the same point. AAMVA is the association of North American motor vehicle administrators. Its guidance on plate recognition says “the constant introduction of new and varied license plate designs greatly expands the types of license plates that ALPR systems are required to process”.
The same guidance explains why some plates read badly. Busy backgrounds and graphics that crowd the characters make recognition harder, and fonts matter because a B can be read as an 8 and a Q as an O. AAMVA last updated its License Plate Standard in September 2025. None of this is settled.
So ask any vendor two things: how new layouts reach the model, and how you will know when they have. An engine that tested well last year is reading a different population of plates today. Put update cadence and support in your specification alongside accuracy.
Buyers in this segment have usually been burned by a supplier going quiet. Model updates and premium support are part of what Anyline sells, so dependability is something you can put in the contract.
What the integration takes
Less than most buyers expect. Your developers add the SDK to the app and call it where the plate field sits today. It returns the characters and a confidence score, and your existing code does what it always did with a plate.
Because the reading happens on the phone, there is nothing to provision. No new service, no keys in a network path, no latency budget to negotiate, no infrastructure review.
License plate recognition, proven at over 99% accuracy
Anyline puts license plate recognition into your own app, on any platform, mobile or web, with no new hardware to buy.
Common questions
What app can I use to read license plates on my phone?
That depends on whether you use software or build it. Parking teams read plates inside the enforcement app their vendor supplies, so the app is whichever one they already run. Vendors building that app add a reader component instead of installing a finished product. Anyline supplies the component. So if you run a lot, ask your software provider what their app uses.
What are the downsides of ALPR technology?
Three are worth planning for. A camera in a drive lane cannot read a plate facing away from it, so cars reversed into bays go unread unless the reader can also capture a VIN. Plate designs change every year, so accuracy drifts unless the model is kept current. And many readers run on one phone platform only, which limits the hardware your team can carry.
How much does an ALPR system cost?
The approach decides the bill far more than the software does. Fixed installations pay for cameras, mounting, power, and cabling on every single lane, which is where most of the money goes. Vehicle-mounted systems pay for hardware and a vehicle. A smartphone reader runs on phones your team already owns, so you pay for licensing and a short integration project.
Does license plate reading work on both iPhone and Android?
It depends on the reader inside the software, and this is where products differ most. Anyline is built natively for iOS and Android, supports React Native, Cordova, and Flutter, and has a web SDK for JavaScript. Many readers sold in this market are Android-only. That forces a vendor to standardize the whole team on Android or ship an app with a gap.
Can a license plate reader identify a car with no visible plate?
Only if it can also read a vehicle identification number. A car reversed into a bay in a rear-plate-only state shows no plate to the aisle at all, and the same goes for a bent, dirty, missing or paper-tagged plate. Anyline reads the VIN from the windshield, door sticker or chassis using the same SDK, so the vehicle is still identified.
Can license plate reading work without an internet connection?
Yes, when the reading happens on the phone. Anyline processes plate data entirely on-device and runs fully offline, so a read finishes on a lower garage level with no signal. Cloud readers send every frame to a server and stop working when the connection drops. If your officers work inside concrete structures, this decides whether the app is usable there.
What accuracy should you expect from a license plate reader?
Anyline reads plates with over 99% accuracy. Treat any vendor figure as a starting point. What matters is the plates your officers see day to day: bad angles, low light, dirt, and the specialty designs your state keeps adding. So ask two things. What was the number measured on, and does every read return a confidence score? A score lets your app flag a doubtful read for a human instead of filing it silently.
