Would it not be possible to use existing image processing and recognition techniques and software to automate the labeling and identification of objects such as houses, instead of outsourcing the problem? asked 20 May '15, 11:06 mm5110 |
That's a fairly broad discussion question, and might not be best suited to the structure of this help site. Usually the answer to "wouldn't it be a good idea if OSM did X?" is "why don't you do a proof of concept of X". Note that there's at least one related GSoC project: http://www.google-melange.com/gsoc/projects/list/google/gsoc2015 (search for "openstreetmap") answered 20 May '15, 11:39 SomeoneElse ♦ |
I'm not quite sure what you mean by "outsourcing". From an OSM perspective, if at all, doing it mechanically would be "outsourcing".