Read a licence plate from a photo.
Drop in a vehicle image. The detector locates the plate, crops it and returns the characters in a single request. This page runs on the demo endpoint: no key needed, ten requests a minute per visitor.
Source image
JPEG, PNG or WebP · up to 10 MBDetected plate
Waiting for an imageThe cropped plate appears here
No reading yet
Choose an image to begin.
The demo is rate limited per visitor and intended for evaluation. Production systems call POST /api/detect/ with a tenant API key. See pricing or request a key.
One request, three stages.
Detection and recognition run as one pipeline on the server. You send an image and receive the reading together with the plate crop that produced it.
Detect
A YOLOv5 detector trained on plate imagery locates the licence plate in the frame and returns its bounding box.
Crop and normalise
The plate region is cropped and resized to the recogniser's input geometry, so lighting and framing differences matter less.
Read
A convolutional recurrent network with CTC decoding reads the characters and returns them as plain text, alongside the crop for verification.
Built for production systems.
The same pipeline behind this demo is available as an authenticated HTTP endpoint. Send a multipart image with your tenant key in the X-API-Key header and receive the reading plus a base64 JPEG of the crop.
- Per-tenant keys. Every request is attributed to your tenant, ready for usage reporting.
- Predictable errors. 401 when the key is missing, 403 when it is invalid, 422 when no plate is found.
- Fast to integrate. One POST, one JSON response. No SDK required; any HTTP client works.
- Deploy your way. Hosted API, or a containerised service inside your own environment on request.
Request
curl -X POST https://<your-host>/api/detect/ \
-H "X-API-Key: $LP_OCR_API_KEY" \
-F "file=@vehicle.jpg"Response
{
"recognition_result": "rk457as",
"cropped_image_base64": "/9j/4AAQSkZJRg…"
}Tuned to your environment.
Need higher accuracy for your environment? DEVIT Technology can train and fine-tune a dedicated model using your approved image data, plate formats, camera positions, and operating conditions.
- Bespoke model training
- Detector and recogniser fine-tuned on your approved images, with labelling support where you need it. You keep control of the data and where it is processed.
- Regional plate formats
- Character sets, layouts, fonts and dimensions optimised for the jurisdictions you operate in, including mixed fleets that cross borders.
- Your cameras and lighting
- Adapted to your camera positions, resolutions and viewing angles, and to the lighting you actually get: night, glare, rain and headlight bloom.
- Your operating environment
- Car parks, access gates, weighbridges, toll lanes or fleet yards: the model is evaluated and tuned on footage from the places it will run.
Indicative pricing.
Monthly tiers for the hosted API, priced per image submitted. All pricing is indicative and subject to a commercial quote. Additional usage and bespoke training are quoted separately.
Evaluation
Free
100 requests
Non-production use
- Full API response
- Email support
Starter
$19 / month
10,000 requests
About $1.90 per 1,000
- Production key
- Email support
GrowthPopular
$79 / month
100,000 requests
About $0.79 per 1,000
- Production key
- Priority email support
Scale
$249 / month
500,000 requests
About $0.50 per 1,000
- Production key
- Priority support
Enterprise
Custom
Custom volume
Quoted per engagement
- SLA and support terms
- Deployment options
- Bespoke model
Prices in USD. Evaluation keys are for non-production use only. Overage, bespoke training, on-premise deployment and support agreements are quoted separately.
Contact us for an API keyProduction access starts with an API key from DEVIT Technology.
Keys are issued per tenant and are never embedded in this page. Tell us about your volumes, plate regions and cameras and we will set up an evaluation key and a quote.