We’re a small team based out of Hyderabad, India. We will be doing a small pilot on Pedestrian routing in Cambridge, MA.
Please refer to our wiki [https://wiki.openstreetmap.org/wiki/NextBillion.AI-OSM] and github [https://github.com/NextBillionAI-OSM/OSM/issues/1] for more details or reach out to us at email@example.com
Why Cambridge, MA? Will people on the ground validate the edits? Does your process identify curb ramps types and tactile_pads?
Have you discussed this with the local community in Cambridge,MA?
As a person who has added hundreds of sidewalk in Skagit County, using AI seems abit of a stretch. To much leaf cover and building overhangs to get good imagery for many locations. Unless someone has driven the area for Mapilary and OSC recently, the Bing street side imagery is really outdated.
I also think that its worth mentioning https://wiki.openstreetmap.org/wiki/Import/Guidelines. If you’re going to import anything into OSM you’ll need to go through that process first.
I’d also suggest that each of you try and get a bit more experience of “normal editing” of OSM first - clicking through from your wiki page finds people who have no more than a couple of dozen edits each.
Thanks for the quick response. We want to start with the US, considering the existing data is well organized and there is decent mapillary coverage. As we could see pedestrian attributes in Cambridge are half mapped already, we decided to contribute to this area. Our current scope of work is limited to adding footways, left/right walkways, crossings and steps and resources would be aerial imagery and mapillary. We would leave a ‘fix me’ tag if local knowledge is required.As of now we resort to manual editing without using AI. From our initial analysis, we have identified Maxar is updated imagery and would be our prime resource to map with. However, Bing and Mapbox have more clear imagery and we would use all three resources and Mapillary to map accordingly. Please suggest
Thank you for sharing the import guidelines :). However, we would want to resort to manual editing of pedestrian routes in a limited area to gain more experience
Hi team! Neat idea.
I would caution against Cambridge, MA, as they already have a very detailed datasets in their municipal files in need of an import process, and it’s a very high profile community locally for government work dependant upon OSM exports. Most of the City’s data is available on their GitHub: https://github.com/cambridgegis.
I’d recommend starting someplace with less on-the-ground resources that still has heavy pedestrian activity, perhaps the City of Lynn just north of Boston, or the City of Brockton to the south? Both are small cities with a broad array of street/pedestrian way typologies, so they would still be good for learning. And neither have current local mapping communities or make use of OSM exports as frequently.
Additionally, I’d recommend reaching out to the municipality and/or regional planning agency for whichever entity you decide to work on, to provide an opportunity for ground-truthing. The regional planning agencies in particular tend to be very helpful, and are looking for additional opportunities to work that do not violate the Commonwealth’s current social distancing policies.
Thank you for some really valuable suggestions. We will not be using any of the ML data for pedestrian routing, we will be using mapillary/aerial imagery and start small. If any issues please let us know, we’ll hold back. It’s just a couple of us doing it and small edits in very small areas using complete manual operations. We look forward to your support