Computer Vision Google Maps project

I need someone with experience of computer vision and Google Maps to train a model to pick objects from a satellite images and provide me the street address of the object

Im open to various approaches but essentially all we want to see is a deployed model that uses either GoogleMaps or open Maps images to run inference on cities and produce images with segmented objects and their addresses

Doesnt need to be an app, could run as serverless code and just send the address and snap shot of the image to zapier for processing

We have a couple of open source projects which might form the basis of this work

As a condition of bidding you agree to keep the resulting code confidential and you agree to assign any code/license/copyright developed during this process to us