Ask HN: How to Build Your Own (Mini) Google Street View?
Hey all, I'm pro-bono consulting with a small European economic area and found myself thinking: "There must be someone on HN who has solved this problem."
Question: With a small budget but ample manpower how would you build a small scale Google Street View to take a snapshot of the road system and infrastructure in an area?
It seems that quotes we're receiving for GPS tagged location data (500k+) are outrageous compared to the cost of buying a camera (~20-30k) and mounting it on a truck to survey the area... then building a small dashboard to have locals tag the data.
In total there are ~300km of roads to survey.
Anyone familiar with this?
Any open source tools for managing this sort of data?
Is it crazy to think we could pull this off for a fraction of the cost?
11 comments
[ 2.8 ms ] story [ 36.4 ms ] threadAlso https://www.mapillary.com/
This was also posted on HN recently: https://jakecoppinger.com/2022/12/creating-aerial-imagery-wi...
https://blog.mapillary.com/news/2020/06/18/Mapillary-joins-F...
Operated now by Grab: https://engineering.grab.com/kartacam-powers-grabmaps
Disclaimer: I work for Grab (but not in the mapping team)
If I would do it again I would dump the images into Photo Sphere Viewer and make a mini-map.
It depend what is needed, but it is conceivable to use a cheap 360 camera on a stick and log the journey with OsmAnd/whatever. If only some images/locations need to be high detail, use a normal camera for those locations. With care the images could be merged, but it's probably not worth it. Just link to it from the photo sphere.
https://photo-sphere-viewer.js.org/
Buy an Insta 360, mount it to a car with a selfie stick. Drive around, set the camera to take a snapshot once a second. Dump images to S3, extract exif to DB, display on top of an open streets map.
You'll have to navigate using clicking on a map. With a bit of extra work, create a linked list of the images, then display buttons on each image for prev/next.
Get in touch, let's talk.
Stay away from the home grown solutions, they are not robust, reliable, or scalable. You'll waste a ton of time and money on them with bad results.
Our systems have been used all over the world to map entire countries with no hardware failures.
Happy to talk!