License Plate Recognition (LPR)

Overview

Meraki MV License Plate Recognition (LPR) is an advanced feature that integrates object detection and optical character recognition (OCR) to automatically identify, read, and catalog license plates from the video stream. This processing happens entirely on the MV53X smart camera (“at the edge”), enabling real-time vehicle identification without requiring external servers or cloud processing.

Recognized plate data is published to an MQTT broker for seamless integration with third-party systems, such as access control, parking management, or traffic analytics.

Intended Use & Key Use Cases

Meraki MV LPR automates vehicle identification for enhanced security and operational efficiency. It is ideal for daytime or well-lit environments with stopped or slow-moving vehicles (up to 15 mph).

Key applications include:

Automated Access Control: Simplify security by automatically identifying vehicles for entry and exit at locations like gated residential communities, corporate offices, or parking facilities.
Parking Management: Monitor vehicles in parking lots for capacity management and track entry/exit times in paid parking facilities.
Vehicle Activity Monitoring: Keep track of vehicle movements in areas such as school parking lots or retail drive-thrus.
Traffic Flow Improvement: Enhance traffic flow and security in gated communities or on private campuses.

Prerequisites

Camera: Meraki MV53X Bullet Camera
Firmware: MV 7.1.2+
MQTT: Properly configured broker and MV Sense license
Environment: North American plates are supported in Early Access

MQTT Quick Guide

MQTT is a crucial component for Meraki MV LPR's integration capabilities, so here's a brief overview of what you need to know.

What is MQTT?

MQTT (Message Queuing Telemetry Transport) is a lightweight messaging protocol designed for IoT devices and applications. It allows devices to communicate efficiently, especially in environments with limited bandwidth or unreliable connections.

Why Meraki MV LPR Uses MQTT

The Meraki MV53X camera uses MQTT to publish detected license plate data. This enables seamless integration with third-party systems like access control or parking management software. MQTT acts as a central hub for this data, simplifying integrations and making them more scalable.

Key MQTT Concepts

MQTT operates on a "publish/subscribe" (pub/sub) model, which means devices don't communicate directly. Instead, they interact via a central MQTT Broker.

MQTT Broker: This central server receives all messages. It acts as a traffic controller, filtering and forwarding messages to interested parties.
Clients:
- Publishers: Clients that send messages to the broker. Your MV53X camera is a publisher, sending license plate data.
- Subscribers: Clients that receive messages from the broker by expressing interest in specific topics. Your integrated systems would be subscribers, listening for LPR data.
Topics: These are hierarchical strings (e.g., /merakimv/<serial>/lpr_detections) that categorize messages. Publishers send messages to specific topics, and subscribers tell the broker which topics they want to receive messages from.

Set Up Your MV53X

Proper camera installation is critical for accurate LPR performance. The following guidelines apply to the MV53X camera.

Setting	Recommendation
Height	Mount the camera at approximately 3 meters (9.8 feet).
Angle	The vertical and horizontal angle to the license plate should not exceed 30°. For vehicles moving near 20 MPH, this angle should be below 20°.
Distance and Zoom	Adjust the optical zoom so that passing vehicles occupy a majority of the frame. At maximum zoom, the optimal detection distance for U.S. license plates is about 16 meters (52 feet).
Focus	Use the camera's focus controls to ensure plates are sharp and clear at the expected detection distance. The text on the plate should be at least 20 pixels high in the image.
Lighting	Avoid positioning the camera where it will face direct glare from sunrise or sunset. The MV53X's built-in IR illumination supports night vision up to 50m (164 ft).

Enable LPR on Dashboard

Navigate to Organization > Early Access > Opt into License Plate Recognition.
Once opted in, you will be able to enable LPR in dashboard by going to Cameras > Cameras sub-tab > The MV53X you want to configure > Settings > Sense > Enable Sense API if you haven’t already > Scroll down and enable LPR configuration > Scroll up and click Save.
Configure LPR options will now appear under the LPR configuration section. Click on it to configure additional parameters for LPR.

Configure LPR Settings

Detection filtering & region of interest

Plate size filters: These fields define the minimum and maximum size (width and height in pixels) of a bounding box for it to be considered a potential license plate. This is a powerful tool for eliminating irrelevant detections. If you are getting false detections on small rectangular objects (like bumper stickers) or very large ones (like road signs), adjust these boundaries to exclude them.

Data output (MQTT)

Base MQTT topic: The LPR feature uses standard topics (e.g. /merakimv). This setting allows you to add your own custom label to the beginning of these standard topics (e.g. <your label>/merakimv). This helps you to logically organize data from various cameras or locations, streamline integration with your existing systems, and manage permissions more granularly within your MQTT broker.

Example Payload:
{
"strings":
[
{
"confidence": 98,
"plate_base64": "E89R2ZYLSwiY8IeSMkO2xPjkDLuPbHBpHuE0EUNXX+hLEQ==",
"string": "03DLUC",
"top_strings":
[
"03DLUC",
"030LUC",
"030UUC"
],
"track_id": 5,
"type": "license_plate",
"vehicle_base64": "/9j/4AAQSkZJRgABAQAAAQABAAD/2wBD...",
"vehicle_id": 31,
"x0": 2850.0,
"x1": 2900.0,
"y0": -7.23,
"y1": 24.7
}
],
"ts": 1738578203963
}
Detection Payload: This feed provides a raw, high-frequency stream of data, publishing a new message for every single video frame where a license plate is identified. It offers the most granular, moment-by-moment detail, making it suitable for advanced, real-time analytical applications that require precise, frame-level insights into plate visibility. However, due to its continuous nature, it generates a very substantial volume of data, which can quickly saturate and overwhelm an MQTT broker and downstream systems if not specifically designed to handle such throughput.
- Plate cropped image example:
- Vehicle cropped image example:
Event Payload: This is a smarter, consolidated feed designed for practical, event-driven integrations. Instead of reporting every frame, it intelligently tracks a unique license plate as a single, cohesive "event" throughout a vehicle's presence in the camera's view, providing a clean and actionable record of each vehicle's pass. This payload is recommended for most use cases because it significantly reduces the data volume and simplifies integration with typical access control or parking management systems.
Event Timing: This lets you control how frequently event messages are sent for each tracked license plate. By setting a minimum time between messages and a maximum number of messages per plate, you can receive useful periodic updates and prevent your system from being flooded with unnecessary events.

Image & data formatting

Image Padding: This setting allows you to customize how much extra space is added around cropped plate and vehicle images. By adjusting the padding for each image type, you can fine‑tune how much surrounding context is included beyond the detected area.

Show Objects Feature

This feature provides historical analytics of license plate readings. It allows you to review detected plates and their confidence levels from recorded video. This functionality is available for historical records only and does not apply to live video streams.

How to Enable Show Objects

Go to Organization > Cameras > The MV53X you want to configure > Click Show Objects > Check Show details.