The solution consists of two QCAM 5 cameras placed in such a way that the passengers in the front and rear seats are recognized. One camera is positioned at an angle to the carpool lane and the other camera is positioned perpendicular to the carpool lane. This arrangement of both cameras allows a general view of the vehicle interior.
The special feature of this solution is that the cameras are equipped with polarization filters specially developed for this application, so that it is possible to see into the vehicle interior even in difficult lighting conditions. In this way, for example, obstructive light reflections can be avoided.
The AI embedded in the camera recognizes the passengers in a vehicle thanks to object definition. Our solution counts the passengers only once by tracking the passengers on all images and merging the data from both cameras according to our highly complex "data fusion" algorithm. This avoids double counting of passengers, as the solution is able to detect the exact position of each passenger in the vehicle.
