Hi all, I’m an intern who is a part of an NGO and as part of that we would like to do some automation changesets around the world identifying trees. All I know is that we need to create a new page on the OSM wiki. Am I on the write track and can you all give me any suggestions or approvals since I also read that I need approval from the community first.
Welcome to OpenStreetMap (OSM)!
If you have not done so already, you should read this:
Automated and semi-automated edits require a high level of skill and a lot of experience. Therefore, you should start by making regular manual edits to OSM to get familiar with the concepts and tools.
Other efforts to do so have not met community standards. I think it is too ambitious to automate this process on a global scale, because:
- available aerial imagery and other sources are too poor in many places,
- some mappers will oppose it no matter how good it might be,
- many trees can be better mapped as forest polygons or tree rows,
- it will be very hard to avoid duplication of data and
- the import will be out-of-date almost instantly, which creates a lot of work for other mappers who do their best to keep the map up-to-date.
Thanks for the reply. I was thinking the same thing. The thing also is that our advisor/boss is very ambitious and wants to push the capabilities of things. I’ll try discussing it with him.
Makes me think of a discussion in one of the community forums. The guy said he had very detailed data taxonomy, species what not but I tuned out.
18 days later, the only photo you get, not needle, not leave, majestically and stoically stands there and was in bloom, on a square on the main road up from Scafa to Sant’Eufemia in San Valentino in Abruzzo Citeriio. Trunk at foot 70-75cm diameter, 3D imaged in Streetview and StreetSide, likely caught in Kartaview and Mapillary, a tree providing dense shade on hot days like last week, August 15 and I’m not an arborist, a treehugger for sure, the old man said it was there when he was a child so lots and lots of CO2 safely stored.
Interestingly, someone has just posted this on the OSM-talk list. Maybe you know each other?
If you manage to resolve the criticisms of our community, you can start asking your boss for a pay raise, because the OSM community can be really pedantic at times (and for good reason, I don’t mean this in a negative way).
For example, some other potential issues with automated processes are that tight groups of trees are in many cases difficult to identify with automated processes that look at tree crowns and the centre of a tree crown is not necessarily at the same XY coordinate as the base of a tree. See for example this tree in Venice that I photographed yesterday.
Hello! Yes it seems like I sent that email even though my colleague here already posted it. I missed his posted. Sorry about this. Instead, I will simply add additional details that I put in the email here to add more context to this discussion: "I am planning to run the script every 30 minutes to process about 100 trees at a time, creating 7 changests for every run/30 min interval. Hence, every 30 minutes 7 changesets will be created, each having about 14 trees/nodes. These trees will be relatively close around a central GPS point. The comment would be similar to “Adding 14 trees all clustered around X GPS coordinate.”
Thank you for the tree crown example you gave. The way we take the GPS coordinates is that we would ask the users to stand at the base of a tree or stand as close to the base of the tree as possible. Hopefully, that will lead to better XY coordinates that refer to the base, not crown.
Does that mean, you will send out users to every tree? This way you would get the approval very quickly. But it sounds to me like the opposite of your previous post.
Yes users have to take pictures of every tree they want to upload. The images will also be added into OSM. However, we need a full picture of a tree. It will be hard to enforce users to stand at the base of a tree and get a good picture. At that point, we hope our users are standing as close to the tree as possible when taking the picture.
We need to see the script, and some sample output (i.e. .osm file) before this all starts.
What validation have you done on the accuracy of the tree locations? While getting as close to the trunk as possible puts the receiver close to the location you want to capture, it also means that some of the signal could be blocked by the trunk (a thick piece of wood with a high moisture content - assuming the tree is alive). The canopy can also block GPS signals if wet with rain or dew.
How do you plan on doing this? OSM doesn’t contain images per se. You could load them into a third party site and link the object in OSM to them, is that what you are intending?
That’s not so automatic after all. You’re mostly providing your users with a way to add trees to the map one by one. That sounds like a very thorough verification process to me.
It is still subject to Automated Edits code of conduct - OpenStreetMap Wiki
which mentions, “other scripted changes made to the database”
Maybe, but we don’t know who these users are. Have they been trained and vetted, or are they people that just happen to have downloaded the app? We really need to see the data so we can do an assessment of the quality.
Why 14 trees at the time? Are you going to create 14-tree groups randomly in a forest?
Somewhere you mentioned you want the users to take a photo of each tree and then take the position of the photo as position of the tree. This will add a considerable error to the position. Rough estimate: position error = height of the tree.
Apart form the formal aspects, please explain what you want to achieve qualitatively as well as quantitatively.
The changes don’t seem very scripted if each tree is individually reviewed.
Sorry, but it will be impossible to “get a good picture while standing at the base of the trunk”, unless you want a picture taken vertically upwards!
e.g. Friendly-Ghost’s photo of the tree in Venice ^ would be taken ~20m (?) away from the tree itself. Unless all of your contributors are going to take a photo, then go over to the tree to record the position, I can see a lot of bad positions being recorded
How are you planning to conflate your data with trees that already exist in OSM? What assurances can you make that conflation is accurate, given the concerns already raised regarding tree position vs. photograph capture position?
He is using a phython-script, so it ist still scripting. One can say: “It’s more in the direction of building a new editor, than preparing a huge automated Edit.” None the less it is an automated or half automated Edit.
From the standpoint as to whether this is subject to the Automated Edits code of Conduct, whether the features were originally individually reviewed doesn’t matter, What matters is that they now have a collection of data, which they want to bulk upload to OSM. It is no different if your local municipality had collected points for all of their fire hydrants by sending someone out with a GPS and camera and then later some OSM mapper wanted to upload that data to OSM.
Now, in terms of evaluating whether an import should go forward, it is good to know how the data was collected, and individual GPS verification is better than some other methods.