I have backed up pictures for users with a small amount of photos (less than 5000) but i struggle to find a good automated what to obtain the full list of sequences id of a user. i still need to investigate the mapillary graphql stuff, or some way to scrap that with Selenium or something similar. if any of you has experience with that.
That was the same problem I ran into. There was no api to obtain those (unless someone can reverse engineer how the website does it).
But there was a user who was going to delete their entire account but I exported all of it before they could. The way I did it back then was just by going to the user’s profile, scrolling as much as possible to load every single sequence and then extract it from the source of the website iirc. it’s very janky. but yeah.
An automated solution would be better
I’m sorry if this is too off-topic, but what kind of server system are you using? I understand that running a server at home and opening your network to the Internet is considered a very bad idea. Accessing a VPN requires admin rights so is not always an option. Or is it nowadays more common to have shell access to a rented server?
Whatever you want to expose to the internet, don’t allow it to connect to the rest of your network. The easiest way to do this is by having it in a Virtual machine that’s limited to a VLAN so it can only access the internet and that’s it.
If you need to access it from your pc you can always temp change the vlan of your pc to that one OR ideally use something like Tailscale to access it.
You might want to also hide the IP of your home (not necessarily for privacy reasons, but just to prevent people from knowing your ip and potentially ddosing it. I’m anxious about stuff like that) so ideally you have a VPS in the cloud somewhere that you route your traffic through.
My plan is probably to expose my server through Tailscale funnel (similar to cloudlfare tunnels) which does that second step automatically without having to get a vps and set it up. Sure, I’m limited by the speed of Tailscale’s servers, but the best home/business internet upload speed I can get in my building is only 60 mbps anyway (for the time being).
Hi ! I have a script to download images from Mapillary, but you need to give the sequence ids to it. If you have a way to get this sequence list, it could be the final step to automate the takeout.
Yeah I think it might have been your script that i’ve used before.
I’m trying to analyze how the mapillary site is fetching the sequence id’s from a given user.
I’m onto something, but it’s not working quite yet. will post when I know more
BTW: to anyone from Mapillary reading this:
Hi, hello. My personal stance on this thing is “we just want to secure our images in case your division all of a sudden gets shut down by Facebook. And it gives us more autonomy over our own data. I personally do not plan, nor would I recommend anyone, to actively remove images from your service once we start exporting images.
Mapillary still offers useful features (and probably has some cool stuff in the works) that Panoramax and such can’t do or won’t do for a while. So I think it’s still beneficial to have images on Mapillary.”
And to anyone else reading my stance: “You’re not gaining much by deleting your images. If anything, Mapillary still provides useful things using those images. Plus they might be tagged in Openstreetmap. So I personally agree with exporting, but not with deleting them after the export”
I haven’t looked into hosting Panoramax yet but bear in mind that pictures take a lot of space and hosting a instance, especially remotely can become expensive so it’s not exactly as “trivial” as Mastodon (it also involves maintenance if slightly)
Oh I’m very aware. i’ve made price estimations and such.
It’s quite the thing.
(hence why i’m scared mapillary might some day be dumped by Meta cause there is no way that division is profitable. (obv it can’t be profitable, it doesn’t make money. But I mean “does the value it delivers within meta make it worth the ginormous cost of the infra”))
The server at my house:
This is separate from the final production OSMBE server.
i finally some hacky way to get the séquences id of someone. i détailled it here, it is a user script to copy and paste on the console of the devtools of your Browser, it clicks and copies séquences id in an array.
I also asked the mapillary forum to follow the GDPR which requires any platforms with user data to give them a simple way to export it in their account.
if you want to push on this feature too, it is required since no less than 8 years in Europa.
This is awesome. I’ll have to play around with hosting an instance some time. I it’d be a game changer if someone forked karta view and made a Panoramax android app.
Panoramax is better (by that I mean: it’s designed in a very modular way. Which Kartaview most likely isnt). For example: let’s say you don’t like the blurring system that Panoramax uses by default. They’ve made it very easy for you to just pick another system instead. Also: when they designed Panoramax, they wanted to use as many different existing standards as possible. So the way to request images from a given area is actually based on an existing standard (usually meant for Aerial imagery). So they didn’t invent the wheel again.
indeed if anyone knows how to do that, (i do not develop android stuff) the source of the kartaview app is here:
Maybe we should start an other topic on this.
Right now i contribute to panoramax using a gopro 360, and my android phone using Open Camera with “rafale” mode + “delay of 1 second” + gps data saved in picture. but it is not ideal.
back in here, @waldyrious i do not find your user on mapillary, can you give an URL ?
Most of the people who capture for OSMBE don’t use any companion apps at all. I never really saw the point either tbh. Just use the built in camera’s timelapse mode (with the built in gps)
(and optionally record a gpx track on your phone if you really want to track where you’ve been)
Obv if it’s a camera without gps, then sure, I can understand.
that is not ideal because i use it as a phone inside a car, and every time you stop somewhere you will have many captures of the same place. Long stops happens a lot in cities, which make the capture mostly noise. There are scripts to get rid of this, but having it done by an app is less workload on the people end, and more on the machines, as it is expected to be.
And less technichal and hacky solutions are better to reach more people, and not only having developers contributing in commons stuff, which is better for inclusion
I feel like I should mention: the mapillary desktop uploader automatically discards images takes after each other on the same location.
Panoramax could do that as well when it’s implemented (if it’s not already).
So there isn’t much point to trying to make hacks to do this ‘manually automatically’ when it already happens automatically. (again, don’t know about Panoramax uploader atm)
From a consumption side I am wondering about duplicated images across services, once a lot of data is exported and imported.
E.g. I am imagining looking at an editor like josm or id, displaying images from kartaview, mapillary and panoramax to get a better coverage. But then seeing the same image 2 or 3 times.
So in panoramax is there any data field existing or planned that would identify images exported from another service (e.g. “mapillary id”). So that a viewer / editor could offer a toggle option to hide those?
so it’s not exactly as “trivial” as Mastodon (it also involves maintenance if slightly)
