Vision-based mobile application for supporting the user in the vehicle operation

In this paper the automatic vision-based mobile application for supporting the vehicle user is proposed. The system, using the image taken with the typical smartphone camera, recognizes three types of information located on the vehicle body: text data, displayed dashboard icons, and components in the engine compartment. The modern programming tools and recognition algorithms, including deep neural networks, were used in the implementation regarded to the limited efficiency of mobile devices. Preliminary tests using prepared database indicate on a high recognition performance. Recognition results and instructions are presented to the user in an accessible way using techniques of the augmented reality.