Toll roads are used by millions of drivers around the world, making routes faster, safer and more accessible. The big disadvantage of this type of road is the control and payment infrastructure it needs. Bearing in mind that the goal is to make life easier and more convenient for drivers, payment systems need to be optimized and automated to ensure the time invested by users directly increases enhances their satisfaction.
Furthermore, a system which comprehensively controls the passage of vehicles and classifies them is an important asset for the company. The information compiled from automated vehicle control is very valuable when making a range of decisions on highway maintenance or abnormal vehicle behavior that may indicate a hazard on the highway.
Foqum's vehicle counting and classification model is responsible for extracting all this information both for the company and the users.
One of the world's leading companies in infrastructure development needs to count and classify vehicles using their shadow toll and payment toll roads. Although physical measuring systems such as magnetic coils or control arcs can achieve good results, the implementation and maintenance of these types of sensors entails high costs, millions of euros being spent on these systems.
Foqum has developed a Deep Learning computer vision system based on artificial neural networks which analyzes the videos captured by highway cameras in real time, with estimated savings over 95% compared to the implementation of physical systems, a count accuracy over 99%, and a classification accuracy over 95% averaged between the different categories.