Capturing the whole city of Salzburg with 360° imagery for Mapillary/Panoramax (Part 2)
Posted by Negreheb on 27 January 2025 in English. Last updated on 2 February 2025.Hello OSM Community,
I’m still working on capturing 360° images of the city of Salzburg for Mapillary and Panoramax. My goal is to enhance OSM data in Salzburg, especially in areas with notes that could be resolved with better visual context - something 360° images can provide.
Progress as of now
Route Planning Challenges
So, i tried some tools to plan a route efficiently to cover as many streets as possible without overlapping - essentially the Chinese Postman Problem.
- Bikerouter: manual tedious labour involved to cover every street
- everystreet challenge: no GPX download, so useless. No developtment.
- RunEveryStreet: More automated approach, but key features missing and some bugs, and I didn’t get a response to my suggestions unfortunately.
Unfortunately, none of these tools fully met my needs, so i did it more practical.
The Solution I’m Using
OSMAnd’s built-in recording feature. I can record my route in real-time and have the GPX displayed at the same time directly on the map. This way, I can decide on the fly which streets to cover next. It also allows me to choose if i want to cycle a path or highway or not and no router is trying to get me back there. This flexibility worked much better for capturing pictures around Salzburg.
Managing the Photos for Upload (Moving Beyond the Mapillary GUI Uploader)
One of the biggest improvements has been switching from the Mapillary GUI uploader to their CLI tool. The command-line interface is significantly more efficient for handling large image sets. I used a powershell script to split images into smaller folders to avoid the problems i encountered with the GUI uploader, like freezing or crashing when processing too many images. With the CLI, the upload is much more robust.
I still have manually go through the pictures to remove duplicates, such as images taken at a red light or while I was checking my smartphone.
I’ve tried the Panoramax CLI tool, and I’m impressed how good it is. Integrates seamlessy into my workflow. Panoramax is already a strong alternative for hosting and 360° imagery. Every day the develelopment is ongoing, improving the integration in OSM-services like JOSM or Vespucci.
Interesting: Detroit’s Success with Street Coverage
In weeklyOSM 748, it was announced that Detroit has mapped 99.8% of its public roads using 360° imagery. The city’s Street View page showcases their success, and a great blog post by Mapillary highlights how Detroit has benefited from SLI (Street Level Imagery). Their approach demonstrates the incredible potential of using comprehensive street coverage to improve city planning and data accuracy.
Next Steps and Reflections
So, i learned a lot about tools and planning and improvisation. While efficient planning is helpful, the ability to adapt on the fly is just as, if not more, important. Switching from the GUI-Uploader to the CLI-Uploader has been a game-changer, saving me both time and frustration.
Next, I plan to continue expanding my coverage of Salzburg. My goal is to capture the entire city, street by street, and make this imagery accessible to everyone through both Mapillary and Panoramax. Maybe even the city is able to benefit from the SLI.
If you have suggestions for tools, workflows, or anything else that could support this project, please share! Let’s work together to make our maps even better.
Let’s keep mapping - one street at a time!
Discussion
Comment from Koreller on 3 February 2025 at 18:26
I wish you every success in your task !
Comment from mycota on 4 February 2025 at 14:13
How do you mount the camera on your bicycle?
Comment from Negreheb on 4 February 2025 at 14:34
A good point to add in another blogpost :D
But, i simply bought a relatively helmet with a dedicated mounting point for a gopro-mount. I also bought some gopro-mount-extensions, but i haven’t used them yet. I also bought a cap with a sewed on gopro-mount, when i’m walking around.
Pictures of both will be added in the next blogpost.