On Tue, Feb 17, 2015 at 1:33 PM, Harald Kliems <kli...@gmail.com> wrote:
> > > On Tue Feb 17 2015 at 1:16:55 PM Ian Dees <ian.d...@gmail.com> wrote: > >> Are there methods of remote sensing (street-level imagery, data from >> other places on the internet) that could help us with the locality problem? >> > Mapillary[1] seems to have tremendous potential there. They've recently > introduced automatic traffic sign recognition [2] -- no speed limit signs > yet, unfortunately. > Could use a bit of work. It appears to be detecting "Share the Road" signs as "Cycleway Slippery When Wet/Icy" signs. Ideally, it would be able to properly identify and differentiate the 200-or-so unique but nationally accepted signs in the current edition of the US MUTCD's Standard Highway Signs supplement with a reasonably high degree of accuracy given a clear photo, including the ones that have regional variation baked in to the standard (ETC logo signs for toll lanes, so signs indicating Oklahoma PIKEPASS, Texas TxTAG and the east coast's EZPASS are treated the same, for example, and it's not thrown off by bus stop signs that have the operator's logo on 'em if the sign's otherwise compliant). Extra miles if it can handle signs that are notoriously noncompliant to the national standards, like Park & Ride and major city bus stop signs.
_______________________________________________ Talk-us mailing list Talk-us@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk-us
