GeoGuessr recon

Are you a GeoGuessr fan ? The game has been trending on twitch for a few months now and grasped a lot of people into civil engineering practices. The goal is to recognize precise details of cities or countries to guess as precisely as possible where is taken the shown picture.
Lots of city afficionados loved to compete in guessing "where did that lamp come from ?" or "where do we use those electrical outlets ?". Though we thought that a machine could take on the challenge by learning by itself. The game GeoGuessr makes us guess a place in the world from a view in a street, a park or whatever place randomly picked in the game's dataset. We will train our software to recognize pictures and associate details inside it to match with known details from its own dataset.

If you don't know the game yet you should check it out !
GeoGuessr

We need to train the software to be able to recognize matching details on given pictures and link it to known details.
To train this machine, we'll use machine learning algorithms commonly used in Machine Learning :
Neural networks
Image classification
Object detection
DeepGeo: Localization using Neural network AI Learned GeoGuessr Google VS GeoGuessr
Using those methods we'll create a supervised machine learning process, that is to say we have
a defined dataset and knowledge of a number of cities/places and we try to recognize randomly proposed pictures and associate them with known datas.