Core ALPR Technology

Automated License Plate Recognition (ALPR) employs optical character recognition (OCR) technology to automatically read and interpret license plate characters. This process begins as soon as a vehicle enters the camera’s view, triggering video processing algorithms to locate and decode the plate’s characters, which are subsequently relayed to a backend server. This process might seem straightforward, but it presented significant challenges when Inex Technology founders embarked on this journey in the 90s. Initially, they could accurately recognize highly standardized European license plates but achieved limited success with diverse American plates.

The differences between license plates of different countries led us to a distinction between the less technically challenging ANPR (Automated Number Plate Recognition), which is tailored for standardized plates, and the most complex ALPR, which excels in areas like America, the Middle East, and other regions with a variety of license plate designs.

The precision of ALPR technology is critical across various domains. On toll roads, even minor enhancements in ALPR accuracy can lead to substantial increases in revenue. In parking management and security environments, accuracy directly influences operational efficiency and customer satisfaction. Today, our technology boasts remarkable accuracy rates exceeding 99% in some of the most challenging environments.

Our evolution from the initial concept to today’s sophisticated systems reflects a deep commitment to innovation and an understanding of the diverse global landscape of license plate designs. By continuously refining our algorithms and adapting to regional differences, we ensure that our ALPR solutions remain at the cutting edge, offering reliable and efficient license plate recognition capabilities across a wide range of applications.

Top Performance in All Regions of the World

Our technology is operational across more than 20 countries worldwide, with the United States standing as our largest and most challenging market. The US and Middle East diverse license plate designs present distinctive hurdles not commonly encountered in Europe, where plate designs are more standardized.
Achieving and maintaining a leadership position in ALPR accuracy is a dynamic and continuous process. Our efforts are focused on retraining universal ALPR algorithms but also on integrating specific regional enhancements to elevate system accuracy even further. We meticulously address factors unique to local markets, such as syntax variations, the orientation of stacked letters, the introduction of new graphic designs, and the differentiation between characters such as “0” (zero) and “O” (the letter O). Additionally, we consider regional-specific fraud patterns, which further complicate the recognition process.
Our systems are engineered to excel under these conditions, with features designed to optimize exposure, resolution, and multi-frame processing. We also excel in identifying vanity plates, recognizing state-specific plates, and distinguishing between various plate types (such as car, truck, dealer plates, etc.). This comprehensive approach ensures our ALPR technology remains at the forefront, capable of delivering unmatched accuracy and reliability in a market as diverse as the United States.

Toll Roads Drive Algorithm Accuracy for Other Applications

A key driver in achieving high ALPR (Automatic License Plate Recognition) accuracy is its application on toll roads, where transactions often include “ground truth” data. This data allows for the monitoring of system accuracy across numerous camera locations, facilitating rapid neural network retraining and numerical validation of accuracy improvements.

Over the past decade, most major toll road operators have adopted ALPR-based “video tolling” as a supplementary revenue collection method. Even on roads where transponders serve as the primary vehicle identification method, approximately 10% of vehicles experience transponder issues, including absence, malfunction, or misreading. Consequently, ALPR has become the pivotal technology for securing billions of dollars in road toll revenues. Enhancing ALPR accuracy by even a fraction of a percent can significantly reduce operational costs.

To maximize ALPR accuracy and, by extension, revenue collection, several statistical techniques are employed in training the ALPR AI engine. First, cameras must capture high-quality images of all vehicles to ensure enforceability and prevent payment disputes. Second, efforts must be made to minimize recognition errors in plate reading to prevent billing inaccuracies. The recognition algorithm is fine-tuned using a Confidence Level threshold; transactions falling outside this threshold undergo manual review. Employing a highly accurate algorithm reduces the need for human review, thereby lowering the toll operator’s back-office costs.

We continuously update the AI engine with license plate data from various U.S. states and countries worldwide, enhancing accuracy by incorporating local specifics.

Our extensive experience in tolling ALPR applications paves the way for deploying high accuracy ALPR for other applications in parking, access control and others.


Full Range of ALPR Cameras 

Our comprehensive camera lineup is the result of three decades dedicated to refining the accuracy and cost-effectiveness of ALPR technology for a variety of applications. We provide a versatile selection of ALPR cameras, designed for specific applications. To accommodate deployment-specific needs and variations in license plate design, we have developed specialized sub-models optimized for all possible installation and operational conditions. Inex cameras can capture ALPR data from distances as close as 5 ft (1.5 m) to over 80 ft (24 m). The systems are designed to produce crystal-clear images under all lighting conditions, ranging from 0 LUX to direct sunlight exposure. Our cameras operate in extremely cold climates, from -40°F (-40°C), as well as in extremely hot environments, up to 152°F (70°C). Finally, our camera selection is designed to meet a broad spectrum of customer requirements, spanning from advanced, feature-packed options to those tailored for budget-conscious consumers.
Sharing the same core AI-based algorithms, our cameras leverage multispectral imaging for reliable acquisition under varying lighting conditions, including direct sunlight, and of all license plate types, including non-reflective plates. Adjustable features, such as motorized zoom lenses and multiple focal length options, are configured based on roadway design requirements.
Our cameras support triggered, non-triggered, on-demand and hybrid operational modes. Triggered mode activates upon receiving an external signal, while non-triggered mode relies on automatic vehicle detection through video. On-demand mode continuously captures data while only transmitting license plate information upon receiving a trigger signal. Hybrid mode combines these approaches, operating normally in a triggered state, but allowing the camera to be externally triggered to record images even if no vehicles are detected by video.
Our premium IZA800 camera series is specifically engineered for toll road environments, featuring a frame rate adept at capturing the license plates of vehicles at high speeds. With an onboard NPU-equipped processing unit, these cameras handle all image processing internally, streamlining the output of license plate numbers through a web service API. This innovation enhances system efficiency and reliability by reducing the amount of hardware needed at toll plazas.
Designed with system reliability in mind, a standard toll road gantry setup includes two IZA800 cameras aimed in the same direction, incorporating multiple redundancies to safeguard against potential failures. Should an IR or color camera malfunction, the system is built to ensure only minimal impact on performance. Moreover, if a processing unit fails, the system automatically reroutes image processing tasks to a secondary IZA800 unit, maintaining continuous operation without significant loss of functionality. This robust design philosophy ensures that the IZ800 series stands at the forefront of reliability and efficiency in toll road applications.
In parking and access control scenarios, the IZA500 series stands out for its accuracy and ease of installation. With built-in image processing and operational logic, these cameras eliminate the need for external computing units and include the necessary outputs for seamless integration with access control systems.
Our entry-level IZ600 camera delivers excellent performance across many applications operating with an external Inex-provided computing unit, the DPU, with dry contact . A single DPU can support up to four cameras and also functions as a control panel in access control settings. The IZ600 is equipped with a dry relay for vehicle gate operation.
The mobile IZM cameras are mounted on a police or parking enforcement vehicles, to process in real time ALPR events via a backend server. These cameras utilize a DPU for the in-car image processing to minimize network communication traffic.
All our cameras can operate with Inex IZCloud software service for continuous performance monitoring, firmware updates, and other functions, ensuring that the system remains at the forefront of ALPR technology.


Vehicle Imaging beyond License Plates

Our additional camera models are designed for synergetic vehicle imaging applications. 
In toll road applications, the fees frequently depend on the vehicle type and axle count, with trucks often contributing to more than half of the total road revenue. Traditional methods of axle counting, such as ground loops, can be costly and do not provide a visual chain of evidence necessary for justifying toll charges in the event of a dispute. Our advanced multispectral cameras, equipped with on-board video-based algorithms, accurately determine vehicle type and count axles, including lift (drop) axles, as well as the axles of towed vehicles and vehicle carriers, offering a comprehensive solution for toll fee assessment. Another application of our vehicle type, axil count and other car metadata recognition technology is in toll road audit systems, which monitor the performance of the system as a whole to help ensure that revenue losses remain within acceptable limits, safeguarding against potential discrepancies or fraud.
Our technology extends to the innovative application of facial imaging for drivers and passengers. We have developed a specialized camera solution specifically designed for capturing high-quality images through the windshield under any lighting conditions. Utilizing semi-visible near-infrared technology, these images are well suited for both visual inspection and automated facial recognition analysis.
The primary utility of our facial imaging technology lies in enhancing security measures by creating a visual chain of evidence that links the driver to a specific vehicle. This application is particularly valuable in situations requiring traffic violation enforcement or in scenarios where vehicles enter secured areas, such as corporate or government facilities.
With the accuracy of algorithmic facial recognition now reaching impressive levels, it is feasible to verify the identities of both the driver and any passengers, associating them accurately with the vehicle in question. This advancement not only bolsters security protocols but also adds a layer of accountability and safety. For example, in the context of secure facility access, our technology ensures that only authorized personnel can gain entry, significantly reducing the risk of unauthorized access.
The ability to swiftly and accurately identify individuals in a vehicle can expedite investigations and enhance the enforcement of traffic laws. This technology has the potential to transform how security and compliance are managed, offering a sophisticated tool that complements traditional security measures and provides authorities with actionable intelligence to maintain public safety and secure sensitive areas.

Contact us


This will close in 0 seconds