Hello everybody, Neural network based deep learning methods for feature extraction using computer vision are giving concrete results and these are also more and more getting used in the context of maps and OpenStreetMap.
Until this moment the conversation has been pretty much polemic from people either hyping or hating these new techniques of programming with automated statistics. My guess: as OSM we will not be able to completely reject these new developments, and personally, I'm convinced it will also not fundamentally change how OSM works. There are many aspects that we are missing to address in an open, concise and argumentative way. Questions like: Where is the red line of applying machine learning? What are beneficial applications of machine learning for OSM? What ethical implications are having these steps? And for sure, there are open questions on licensing in the context of machine learning. Furthermore, I think it is good to have a place to have technical discussions on how to implement certain tools and talk about engineering and technical details. So far, most conversation has happened in private blog posts, and separated/closed groups. Our communication bubbles are growing, which is not good for OSM, and I would like to see a less segregated and constructive conversation to develop together the OSM way of using machine learning. Therefore I want to invite everybody interested to join this new mailing list, where we can share opinions, insights, experiences and concerns. https://lists.openstreetmap.org/listinfo/machine-learning Thanks, Felix _______________________________________________ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk