An open-source street-level mapping platform. Viable?

I’m sure everyone is aware of the recent acquisition of Mapillary by Facebook. Many of you might also be old enough to remember the Grab acquisition of OpenStreetCam too.

Regardless of your view on Facebook and Mapillary, one thing is certain, it has renewed interest in launching a truly open source alternative: https://forum.mapillary.com/t/mapillary-2-0-opensource/4210

Here at Trek View we’ve been working on tools to process and upload panoramas for some time.

We’re really supportive of Nick Whitelegg’s effort with Open Trail View (aimed at footpaths) and will continue to support it.

We’re also currently working on a web / desktop processor / uploader (for Google Street View, Mapillary, and Open Trail View) called Map the Paths (name to be confirmed). Take a look here: https://docs.google.com/presentation/d/12G3fzBOi_7bRKyLgyk2kzoIQy27FOPW6s6Z2gpspZDc/edit?usp=sharing

I’d love your feedback.

I guess the questions I have about building an open-source street-level mapping platform are:

  • What would be the aim of an open-source Mapillary? Or more generally a street-level mapping platform?
  • Why is it needed (what would it be used for?)? Who would use it?
  • What features would they want / need?
  • Distributing a large amount of images in a geospacial DB can quickly become costly. How could it be funded?
2 Likes

I can’t add much to the conversation, but here’s some reads you might want to consider when planning a solution:

1 Like

Hi Dave,

As you know I’ve been discussing some ideas with one guy in particular from the Mapillary forum. One idea we did agree would be worth exploring is the idea of a distributed system. In other words, rather than spending an inconceivably huge amount of money on the infrastructure necessary to host the entire planet, we have individuals or groups hosting the platform for their own city or county on their own server, AWS cloud space, etc.

Then, when a user requests imagery for a particular location, a ‘master server’ could then locate the server for that specific location and download from there.

We also discussed the need to align the panoramas correctly so that they were all ‘flat’, maybe Mapillary’s own OpenSfM could help here.

You’re already working on PII blurring, of course.

One thing that would be valuable would be o separate out the core engine of OpenTrailView, to make it reusable for different projects. This could include, on the one hand, a possible open source global street viewing platform, or on the other hand a much smaller-scale project such as a national park’s own app to allow potential visitors to explore it. The idea of using OpenStreetMap data to auto-connect panoramas seems to work well, so creating a reusable platform from the core of OTV would, I believe, be worth doing.

Another potential approach for an open source street-viewer would be to run a project to map one city (probably a smallish one with historic interest) and add some value to it not offered by Google (have to think about this) - and then if people like it, who knows, maybe it could attract funding.

Nick

1 Like

My focus is on your last paragraph, I’ve spent a sh*tload of time building an interface upon MapillaryJS. I got the developers to add a few options I needed, I was happy… (excluding the lack of time I had to work on building further on it…) then FB bought them… the distributed structure is the way to go, I am certain of it. We need to formulate a basic structure where those who will want to use it (in our network?) can agree upon. I am thinking about a database format (simply punt, SQL table with fields x, y, z) and a structured way of communication (OpenSource API built upon JSON?)

Can you share some of the examples of the things you have built on MapillaryJS @Eesger ?

Another potential approach for an open source street-viewer would be to run a project to map one city (probably a smallish one with historic interest) and add some value to it not offered by Google (have to think about this) - and then if people like it, who knows, maybe it could attract funding.

I like this idea @nickw – maybe we keep going with the New Forest :slight_smile:

I shared my geo-location-POC to give an impression of what I can develop… it is something I haven’t seen anywhere on the internet (except with MapillaryJS and GoogleMaps, and a few other high price applications)… now with my documented POC (proof of concept) the knowledge is “out in the open”, for everyone to use. I hope it will benefit the community.

I have literately spent hundreds of hours on my “GeoArchive” project, I am getting there, but there still are hundreds of hours that need to be invested into the project.

To be very honest, I do not know you, therefor I am hesitant to share “it all” with you. There is a business concept in there and I need to put my enthusiasm on a leech from time to time to prevent hurting myself. Maybe it is a good idea to (virtually) meet? I am a strong believer that one stands stronger together then alone, providing that those one stands with can be built upon.

My apologies if the above text appears a bit blunt… English is not my native language and maybe it’s a Dutch thing :wink: but I need to “get a feel” with whoom I’m working…

Understood.

We’re pretty open with sharing what we’re working on. It gives us the best validation on what to work on. The thing about the tech world is, and I’ve been in it a long time, is that the barriers to entry are low. At the end of the day, if your idea is good enough, someone, somewhere will steal it. Sometimes, that’s the best form of product validation :slight_smile:

If that’s problematic for you, and I fully understand why, it’s probably best to keep it under your hat. If not a video call is a great idea, drop me an email dgreenwood at the trekview domain and we’ll set something up.

How do any of you plan to compete with the deep pockets of Silicon Valley?

See page 11: https://www.mdpi.com/2220-9964/8/5/232

Facebook has contributed over 800,000 kilometers of mapped roads to OpenStreetMap. They rank third in kilometers mapped behind Mapbox/Development Seed (1.69M) and Apple (1.64M).

I do not want to compete with them?

There is now no good platform that:

  1. respects privacy
  2. has a good & scalable structure to place & process (360) images
  3. last but not least: a user friendly GUI (user interface) to show the images.

Mapillary used to be that one company where I had the impression that it found the mix between Open Source and being a commercial entity.

If there is one thing your list proves is that:
"the deep pockets of Silicon Valley" can contribute to OpenSource…
but the fact that Mapillary was sold to FaceBook proves that:
OpenSource structures are not the same as Open Structures at a commercial company…

Thus I believe:
main building blocks can only provide an enduring contribution to the community when they are OpenSource

That is why I believe that we need a structure where at least point 1 & 2 exist and are interchangeable. With that I mean that everyone should be able to install a “GAOS” instance and connect to an other who has one also if they want, combining the data and so increasing the coverage of both… creating a network… thàt is the power of the community!

GAOS: GeoArchive Open Source, or call it OpenPano or whatever :stuck_out_tongue:

Was it not that OSM was first seen as an impossible idea? One that could never work? But when the Dutch AND (if I’m not mistaken) donated their data to the project it caught momentum… and look where it is now! It’s better then Gmaps AND it’s OpenSource… thàt is what I see for the future of 360 images also. Maps are not the copyright of one company, they should be the property of the community. Same goes for 360 imagery.

PS: I dare to dream :wink:

1 Like

Indeed, that sounds like a good plan! :slight_smile:

My reply above was mainly focussed on my experiences with Mapillary, Others:

gMaps = Google
Privacy, nah…
Closed source? Yea… one can do iframe, and there once was an api I believe… but it is so limited and closed… yèch…

OSC
I must be honest, I looked at that briefly, for usage to perfect maps? cool! API? Mèh…, GUI? blèch… sorry. Also I didn’t get it, it is open, but was bought by a company… I didn’t see the reason why… I didn’t dare to risc it…

So I thought about it, and some more… so I got to the idea/vision written out in a few centences above… the idea has more depth, but to write that all out… I can talk faster :wink:

I love your passion @Eesger.

I think you’re on to something. I really do.

As you note, it’s not about competing with silicon valley. It’s about doing something better than them that solves a real problem – changing course is hard at a big company when compared to a nimble startup.

And if you do want to make it a competition… just look at the number of non-corporate editors vs. big tech co contributors to OSM.

1 Like

Pos 8 of Joe Morrison is one of the things I genuinely hate… I love it that there are people investing in sharing their knowledge… but I must admit that this point has held me back in shearing my efforts in several fields…

I’m working on a project, the subject mentioned is part of it, that uses some great open source projects… if the moment comes that it starts to generate (seriously) more than it cost, I hereby promise that I will give back

The big challenge, getting there :stuck_out_tongue:

Great can’t wait for the Map the Paths software

1 Like

HQ | Blog | Spotted a Trekker? | Become a Trekker | Facebook | Instagram