The Intelligent Camera

Leading the Way in Video Detection

CAMEA has a long history of bringing innovations to life in video detection. In 1995, the company introduced a system for Red Light Enforcement. It was exceptional as it needed no external sensors and was independent of the intersection controller. The set of cameras did all the work, capturing records evaluated by the SW. Over the following quarter century, various Intelligent Transportation Systems were developed, installed and operated by the company. These include Weigh-In-Motion, Spot and Section Speed Enforcement and others.

The new generation of cameras is launched in 2020, exactly 25 years after changing the market for the first time. These all-in-one cameras were developed in cooperation with internationally recognized experts from the Brno University of Technology and a startup company Cognitechna. They will be used for all transportation applications and transform the way the systems are installed.

ITS Sites Before and After

A vehicle entering a section in which its speed is measured is an example of such a site (see Reasons to Apply Section Speed Enforcement). Before the discussed innovation in order to capture the vehicle, the camera installed on a pole must be connected to a processing unit where the record is processed using SW for Section Speed Enforcement.

The intelligent camera is now the only component needed. Edge computing with artificial intelligence – convolution neural networks (CNN) and deep learning allows saving costs on installation of additional HW. All video detection, recognition and evaluation algorithms are integrated in the camera.

The Way to Best Results

The artificial-intelligence algorithms in the camera also enable "training" of the system according to user needs. With a sufficient sample of quality training data, the overall quality of the results can be significantly improved. An example is ANPR (Automatic Number Plate Recognition), also called LPR (License Plate Recognition). For most of the systems, the accuracy of reading is absolutely crucial.

In the Czech Republic, CAMEA solutions reach around 99% reading accuracy. However, these results did not come with the SW when leaving the laboratory. The algorithms were trained by working with real records and the accuracy improved until it reached these outstanding results.

The license plates differ from region to region. Therefore, it is very challenging to reach high reading accuracy for many countries, but at the same time good reading accuracy is crucial for reaching the best results of Intelligent Transportation Systems. CAMEA treats business cases individually, willing to customize as much as possible including the ANPR SW. The best way for the customer to achieve such results is by doing a pilot installation in order to gather records. After a months time or so, the improvement will be significant and the expected results may reach 98+% accuracy.

The state-of-the-art technology used in the intelligent camera is based on an in-house developed low-power hardware. That includes, along with the computer technology, a programmable logic accelerating the most important parts of the embedded digital signal processing and machine learning algorithms, including image classifiers and convolutional neural networks. All that is now available for an affordable price.