Hacker News Reader

Comment by andrewljohnson | Hacker News Reader

Great find, I'll add that to the references on DeepOSM.

Looking at that repo, the main differences are:

* that code applies to a specific (massachusetts) data set, while DeepOSM uses nationwide USA NAIPs trained against OSM extracts

* they seem to have implemented the full Mnih CNN - DeepOSM just uses a single layer as is

* Mnih didn't touch on using the infrared band

In general, that project seems more mature in replicating Mnih than DeepOSM. Our next goal is a website to show errors in OSM data to be corrected, rather than finish making the net deeper.

edit: added links to paper and code https://github.com/trailbehind/DeepOSM/commit/8053057635a1a0...

View on HN

When you say:

> Our next goal is a website to show errors in OSM data to be corrected

Do you mean feeding the data to one of the existing projects that crowdsource fixes based on lists of potential errors, or a new custom site?

Custom site to start at least, I don't consider any of this really production ready yet.

Here's the app work: https://github.com/trailbehind/DeepOSM/compare/feature/deepo...

It might be worth considering integrating with

https://github.com/osmlab/to-fix

https://github.com/maproulette/maproulette2/

to help with evaluating/fixing of the errors you find.

You might reach out to the OSM team about this becoming a core piece of the OSM infrastructure.

Second bit of feedback - http://improve-osm.org/ does a pretty good job of highlighting areas of potentially unmapped areas. Via slack, I've made the suggestion to them to open their platform up to multiple sources such as strava gps traces (via their tile layer).

Would you be open to working with them to deliver the potential missing roads; so that you don't have to reimplement a lot of the basics?

I don't know what I'm open to, this is just something I do for fun, so I don't have any real roadmap besides the next step.

Would you consider something like Strava's aggregated GPS traces to complement how confident you are with detected roads?

For example: http://labs.strava.com/routing-errors/#250/17/-46.42546/-21.... gives a fairly strong indication that something is navigable on bicycle or foot; even if OSM hasn't mapped it yet.

Yeah, I think there is a big future in automated and semi-automated OSM updates, combining trace data and machine learning. That was part of my interest in this.

My company Gaia GPS has millions of GPS traces from hiking and similar outdoorsy activities, so I thought to combine that data (though footways may be more challenging than roads to classify). OSM trace uploads and Strava both seem like good data sources to do this work too.

Reply on news.ycombinator.com