GPS service provider Telanav has released a dataset of over 50,000 street-level images of common road signs to foster the development of machine learning algorithms for map-making. The images are manually annotated and include more than 55,000 signs from over twenty types of signs, including traffic, turn restriction, and speed limit signs. Machine learning algorithms that could detect the features of a sign, such as the name of a street and turning requirements, could quickly update inaccurate maps.
Image: Telenav’s Dataset