ANPR-X AI based license plate recognition
License plate recognition is around for quite some time. In the early days it was only used for law enforcement, deployed by governments and ministries. It was expensive, devices were huge and license plate capture was slow. Nowadays with the technology being evolved and more affordable, it finds it ways in many different applications. It becomes a part of our daily lives. We can enter a parking facility or holiday park without even opening our windows or automatically start a preselected program when we enter a car wash venue. With ANPR-X we take license plate recognition a step further.

ANPR-X = hardware + software

ANPR-X captures more than just a license plate. It gathers comprehensive vehicle information to look beyond license plate recognition. It implements contextual intelligence and additional image analysis to augment ANPR. The result offers all the information required for more fine grained and sophisticated applications.

This thorough video content analysis (VCA) is only possible because of the ongoing development of hardware and further integrated circuits or processors. Along with ANPR-X, AVUTEC introduces the Gatekeeper X-series. A new generation of smart camera’s boosted with neural processing power. Each Gatekeeper X can run multiple recognition algorithms in parallel. The result: a multitasking ANPR-X camera.

ANPR-X: the elements

1. The AVUTEC ANPR engine

The base of ANPR-X is our AI processed ANPR engine. This technology is backed by over 10 years of ANPR research and development. During this time video analysis developed rapidly. Deep learning and machine learning came to the stage and became essential. Based on both computer vision algorithms and deep learning, our engine uses the latest AI technologies to achieve its accurate and fast performance. It features generic ANPR and country modules. Tailormade modules can be trained or configured to meet project requirements.

2. 40° angled recognition

From the very beginning, our license plate recognition engine is trained to read plates from difficult viewing angles, challenging contrasts and low resolution images. The result is an engine that captures license plates at an 40° angle and a resolution as low as 100 pixels per license plate, even at the darkest hours of the day. These qualifications ensure freedom of positioning, saves installation costs and facilitate ease of deployment.

3. Contextual enrichment

The augmentation of ANPR reads with car metadata, directional information and timestamps provides contextual data to trigger more specified actions or alarms. Time-restricted parking or e-charging can be implemented and alarms can be generated to assist law enforcement and security officers in an alarm center.

4. Binary vehicle classification

A vehicle with an unreadable license plate slips through regular ANPR. This gap is now filled by performing binary vehicle classification. It is a routine that knows when a vehicle enters the scene. It answers the binary question: Is there a vehicle or not? A vehicle without a license plate read can then not be missed anymore, making ANPR-X even more accurate as it already was.

5. Multiclass vehicle classification

The distinction between trucks, motors, cars, bicycles and pedestrians on top of ANPR enables refined access control. A truck can be granted access to the perimeter of a logistic center, while an arriving car or pedestrian will not be allowed. Meanwhile the staff receives a message to announce the arrival of the truck and the details of the load. A display leads the truck driver to a specified docking station.
The ability to detect people and vehicles while reading license plates provides a complete overview of traffic flow in a city landscape or business park. All done within one single camera!

rule engine to trigger actions on predefined events
6. Scripting

To complete ANPR-X, scripting provides the last step to tailor an ANPR application to the needs of any project. With scripting, personalized behavior can be implemented. Scripting uses all the information coming from recognition to trigger pre-determined actions.
Scripting opens a barrier or flashes a light, only when conditions are met. It can open a barrier just for white cars or inform a security officer when certain license plates are in scope.

7. Data analysis

The collected data provides a rich source of information to gain insights and statistics about traffic flow or vehicle visits. Recognition results can be displayed using CortexDashboard to visualize results and insights. Anomaly detection algorithms detect deviations and generate alarms derived by comparing stored data with new incoming detection results.

Gatekeeper embedded ANPR camera ceiling mount
8. X-series: embedded ANPR-X camera

Engineered for both ANPR and deep learning tasks, X-series cameras are real all-rounders. A Gatekeeper-X can multitask, perform complex video analysis and communicate results to third party systems simultaneously, without the need for extra server capacity or middleware. The X-series form the new generation of AVUTEC’s smart camera systems, bringing embedded license plate recognition to the next level.

ANPR-X is modular, offering the proper composition and video analytics for each project. In combination with our X-series we are proud to offer ANPR-X, taking embedded license plate camera’s to the next level.
For more information about ANPR-X and the Gatekeeper X-Series contact our sales team at [email protected].