In this talk, Telenav’s lead AI engineer Adrian will take you on a road trip. The starting point is 150 million OpenStreetCam images contributed by mappers and drivers across the globe. On the way, we will make stops at every point of technological interest. We will show how street level images are assessed and segmented, how salient features are extracted, and how these features are compared with existing OSM data. At the end of the journey, we offer improvements to the OSM map, ready to be validated and applied by human mappers. We can show you all this in great detail, because the AI parts of the OpenStreetCam stack are now open source, and we want the OpenStreetMap community to improve and extend it, to make OpenStreetCam more valuable for everyone. To jumpstart this, we organized a competition inviting improvements to the stack earlier this year. To conclude the talk, we will share the results of this competition, and our plans for expanding the types of features we can detect.

Back to program