see prior art
Basic steps are:

1. Convert HRSL to vector
2. Add a buffer (~100-200 m) to cover areas that are adjacent to settlements
3. Intersect the buffer output to admin boundaries (so we can filter by admin jurisdiction later)

• Inventory of current validation/QA tools
• Identify tools we can use for mapping roads
• Create a validation guide for use during task validation

• Post as a wikipage under Philippines in OSM wiki:https://wiki.openstreetmap.org/wiki/Philippines/Tabang-AI
• Announce in talk-ph list
• OSM diary

Siquijor Island, OpenStreetMap Cyclemap Style

Using AI-based detected roads, the goal of this project is to improve road coverage for the whole island of Siquijor.

We will use this ticket to coordinate mapping efforts. If you have questions, please comment on this ticket.

To contribute:

• Visit: https://tasks-assisted.hotosm.org/project/16
• Read the detailed instructions in the project page.

Changeset comment (OSMCha):

Mapping roads in Siquijor, PH. https://github.com/OSMPH/Tabang-AI/issues/10 #osmph #tabangAI

Partner organization: Local government of Lazi, Siquijor through @kinleah

Daghang salamat!

cc @kinleah @govvin @seav

• Create RapiD task ‼️ @maning
• Set Lazi as priority area ‼️ @maning
• Remove Lazi as priority area, as soon as it's complete. ‼️ @maning
  • Contact other LGUs, thru Lazi or Nikki for possible collaboration ‼️ @govvin @kinleah
• Send letter to LGU (LCE thru MPDC) to introduce the initiative and seek collaboration on how to publish local geodata that could be added to the commons, (emergency-related POIs, public amenities, etc, etc) ‼️ @govvin
• Organize remote mapathons
• Validate all tasks
• Organize local socialization

Originally posted by @kinleah in #5 (comment)

Facebook released their country export of their MapWithAI detections, including the Philippines here

To help with visualization and validation, a WMTS layer is made available by the TabangAI project hosted courtesy of Mapbox: https://api.mapbox.com/styles/v1/osmph/ck21d55mx10gu1clclasftm17/wmts?access_token=sk.eyJ1Ijoib3NtcGgiLCJhIjoiY2syMWVqaTR1MTlrNjNtdWcwZ3pseG01eSJ9.c1cgQJX9P8pzQoGXWIgurQ

NOTE: This layer's features was simplified, to reduce the file size, and will have a slightly different geometry with what's actually available from RapiD.

For example, in JOSM just add a new imagery layer with the value wmts:https://api.mapbox.com/styles/v1/osmph/ck21d55mx10gu1clclasftm17/wmts?access_token=sk.eyJ1Ijoib3NtcGgiLCJhIjoiY2syMWVqaTR1MTlrNjNtdWcwZ3pseG01eSJ9.c1cgQJX9P8pzQoGXWIgurQ

We don't have one, let's have one.

I imagine this repo will be mostly derived data and coordination of effort. So more content than software. Any open content license should fit.

@seav @govvin

This tasking manager project is for evaluating the mapping workflow of tasks-assisted.hotosm.org + RapiD.

• Use FB HRSL (with a buffer of 200m) during tasking preparation to identify areas to map.
• Use RapiD to improve the road coverage by:
  • validate roads detected by the tool;
  • add missing roads not detected by RapiD
  • align and improve shapes of existing roads;
  • tag roads accordingly and;
  • fix validation issues ( i.e. overlapping geometries, too many nodes, etc.).

Mapping roads with RapiD in Camiguin Island: https://tasks-assisted.hotosm.org/project/9

After the testing of workflow we would like to answer the following:

• Is it effective to use the HRSL data to prioritize areas to map?
• Does RapiD increased the perceived efficiency of mapping according to the users?
• What are common mapping mistakes we found during validation and how can we avoid them.

@govvin Please lead the first stage of mapping with your CWTS++ students.
@adisadit Thank you for preparing the TM project!

Good day! Im Mikko from Grab and we are interested to try out and help with this new Machine Learning initiative. We hope you can consider us as we would really like to support and understand such innovations. Thank you

During mapping of Siquijor we discovered that some grids were queued for mapping and validation several times.
See screenshots below, darker grids means duplicated polygons.

TM visualization

Actual polygons used for creating the projects

It looks like the grids are duplicated along the edges of municipal boundaries.

Next actions

• Inspect the whole dataset if this is persistent in across
• Re-generate the 500m grids to remove duplicates