The main outcome of the project was the development of a compact and portable camera for automatic vehicle detection and identification – NeuroCar ANPRCam, with built-in image processing (so-called edge computing). The functionality of the device is aligned with the wishes of potential clients, identified based on surveys conducted in the first phase of the project:

• the device independently processes video streams and generates real-time comprehensive information about identified vehicles, including: license plate number, country of origin, vehicle type, make and model, direction of movement, and current speed,
• it detects vehicles that cannot be identified (e.g., without a license plate) and determines their type as well,
• it self-calibrates and can detect disturbances affecting measurements,
• it continuously records for at least 24 hours using its own power sources,
• it enables the anonymization of results in compliance with European data protection regulations,
• it is prepared to cooperate with supervisory infrastructure (BackOffice) operating in the cloud,
• it achieves the accuracy required for planning and monitoring road infrastructure.

In line with these assumptions, two variants of the device were prepared – for measurements on one or two traffic lanes. Functional and quality tests confirmed that the device achieves a very high detection rate (>99%), identification rate (>97%), and vehicle classification rate (90%) under the specified boundary conditions provided in the product documentation.

The first significant, intermediate result of the project was the development of several new functional modules (calibration, adaptation, image quality assessment, service monitoring, fault detection) and the further development of existing core modules (detection, classification, data integration, anonymization) of visual detectors. Currently, all operate based on the latest technologies and optimized computational algorithms. The modules have been implemented in all devices of the NeuroCar product line as the so-called firmware “NeuroCar Terminal Vehicle Identification”, determining the innovative advantage and competitiveness of the product line.

The second significant, intermediate result of the project was the development of supervisory software – “NeuroCar BackOffice”, operating in the cloud (using REST Api/Docker/Kubernetes), allowing real-time monitoring of camera operation, collection of measurement and diagnostic data, and analysis of this data using various analytical modules. The commercialization of the new product and technologies involved:

• the sale of the new “NeuroCar ANPRCam 4” device,
• the service of performing traffic measurements using proprietary measurement devices,
• and the sale of licenses.

The project was carried out in collaboration with the following entities:

• DTV-Verkehrsconsult GmbH (Germany)
• TB Traxler GmbH (Austria)
• Fraunhofer-Institut für Verkehrs- und Infrastruktursysteme IVI (Germany)