Philippines relief map at Rizal Park, Manila

Another approach I do is: 1. Download features from overpass. 2. Load the features to JOSM’s Todo list. 3. Start resolving items while the “Download OSM data continuously” is active. This makes sure you have the latest data each time you fix and item in the todo list.

Sa uulitin Leigh!

❤️🚪

Thanks Ron! I too enjoyed all the scholars’ lightning talks.

This is very cool, looking forward to the next phase!

We had a chance to test RapiD via the https://tasks-assisted.hotosm.org/ at Pista ng Mapa 2019 conference in Dumaguete, Philippines last week. The general comments of the local community is positive and has the potential of improving the speed and quality of mapping (tweet of the demonstration here).

A couple of notes below:

1. Unlike other parts of the world, detections in the Philippines are fewer, this is because most areas (population centers) were already well mapped by the local community. The detections provided by RapiD allowed us to identify the last remaining gaps to complete the road coverage.
2. Geometry quality of detections are very good (although jot perfect) and is comparable to an average mapper. This allows us to focus on tagging quality and connectivity instead of tracing. The time to investigating the correct tag based on local knowledge augmented by imagery will definitely improve the overall quality of the data.
3. Validation is integrated to the workflow. Most common geometry errors can be avoided before upload.

We are continuing the tests and we will share back observations to FB to improve the workflow. Good job to those who built it!

How did you prepare the subset with rs subset to set it up for 80/10/10?

I found something on the net that splits a text file by line based on %, so basically you input your cover csv file to this script: https://gist.github.com/maning/8a381a66e4f245429d67a6f39e205e24

Yay new release! I definitely give the docker build a try!

❤️

@tomas straupis,

Is there a difference in trained model prediction results from epochs 20-50?

Good idea I have not tested the other high IoUs, I’ll report back here of the result.

@NewSource

can you please share the source 660c5321-0334-471f-bca5-829d85fb1d40.tif ?

Sure, here

by the way what the way used to assemble several tiffs together ?

I mentioned it here:

Since I have to do this in all individual imagery, I created a script that loop through each imagery list from OpenAerialMap.

http://oin-astrodigital.s3.amazonaws.com/LC81230392015234LGN00_bands_432.TIF

It looks like you are downloading landsat imagery, is this what you want to download? Depending on the zoom level you download, this is expected to be low resolution (28m pixel resolution).

Following this guide, I got the following results. Hard negative mining helped a lot.

@planemad, @SimonPoole, Thanks! This make sense now.

Re: charts, I use the basic charting of matplotlib, will share code later.

Awesome work! Its so nice to see that we continue remote mapping + fieldwork + local participation in all our OSM training.

Love the bibingka approach!

Great work! The jeepney routes looks really nice.

Hit me if you have questions. Keep on mapping!

@Sunfishtommy , Strava heatmap are available in iD and JOSM as background layer. See this entry in the wiki: http://wiki.openstreetmap.org/wiki/Strava#Data_Permission_-_Allowed_for_tracing.21

I’m not sure why you are assuming that GADM data has a acceptable licence for OSM

@Simon, thanks for the link.
That is partly the reason why I’m saying the license is incompatible (paragraph 4), some borders from selected countries were given explicit permission, others may get this impression that GADM is kosher for OSM. I should have added that link in my post.

It is mostly a little offset that is all.

@Rovastar, that image is actually one of the “best” ;). In other areas it is worse, as Eugene mentioned in his reply, this is not a consistent offset issue. The second image in my post of Burauen, Leyte has a boundary offset by as much as 3km from the farthest village south. The imported boundary of this area closely resembles the GADM data although it might not come directly from GADM. As I said, there maybe other sources but all of these data are similar to GADM in terms of quality.

adding something that is roughly/nearly/high percentage correct instead of the alternative which is, well, nothing at all

In most cases, I agree with the “roughly/nearly/high percentage correct instead of nothing at all” mantra except for admin boundaries. There are various factors in play such as maintaining relations integrity, too few mappers, etc., but more importantly, unlike a road or a building, mappers cannot easily correct them through satellite imagery or field surveys with a GPS.

@Alan Bragg, thanks every node/way contributed helps.

Hey! Really glad another Marikina mapper here! For photomapping, I suggest you also look at mapillary. Lately, I’ve been uploading my Marikina photos in mapillary: http://www.mapillary.com/profile/maning

Interested to meet sometime?