Autonomous, camera-based recognition and interpretation of traffic signs is a relevant requirement for many warning and vehicle safety systems. Providing information about the maximum speed limit, no-passing regulations, no-entry situations, and warning and danger signs supports the driver in complex traffic scenarios and becomes necessary for a complete scene interpretation for autonomous driving systems.

Computer-based traffic sign recognition is based on machine learning and pattern recognition, which relies on classifier trainings with intensive data usage. For this purpose, Continental uses a developed MATLAB^® tool chain that provides automatic learning of different sign types, interacts with different databases, generates synthetic sign samples, generates code, and provides interactive GUIs for operators to monitor and evaluate the classification training.