Model for finding houses on Google Maps?

Hi! New guy here, sorry if my question doesn’t make too much sense.

I’m trying to automate some time-consuming tasks with computer vision and machine learning. For example, given a map of a rural area, do you believe I could use some sort of automation to recognize parts of it corresponding to a human-built structure, like a home or a barn? I’m especially interested in doing it with Elixir and Phoenix LiveView, especially if I can just consume Google’s API instead of working with screenshots.

Probably… There are definitely better places to ask this question though where this question will get more tractions. (There’s gotta be a computer vision or machine learning subreddit where you can reach a wider audience.)

I think that OpenStreetMap uses the same concept as Google Maps (interpreting building shapes from overhead imagery) so maybe you could look into that as well.

1 Like

Thanks for your reply! I’m going to bed now as it is late here but if you could please point me to a specific resource about OSM I’ll read it when I wake up. Thanks in advance!

You replied quick :D, I didn’t get a chance to edit my post with a mention of Bumblebee, which is an Elixir library that allows you to easily integrate with machine learning models available on HuggingFace.

I did a quick poke around on the topic of OSM building recognition (Not trying to be a dink with the Google link, there’s a ton of info there and it’s worth a look), which led to a few links, most notable this one.

From what I’m seeing, there’s a few repos out there, and they all have Python + Jupyter notebooks. Which makes sense since Python is big in the ML community since it’s a great glue language and is used widely in academia.

1 Like