[Week-7 Buildings In The City]

Our project was a photo classification project. We tried to make this classification based on the building objects in the photographs. The classes we used in our classification phase were 7. 7 Class 7 meant the geographical regions in Turkey. What we want to do is to classify buildings that
are shaped according to geographical features. The probability of knowing correctly is 14 percent when you randomly guess what geographical area your building photograph belongs to. We estimated it by 57 percent using machine learning methods and algorithms. We have trained
our system in this direction.

We were hoping to get the best result using CNN. We achieved the best skull in CNN with 57 percent success. But the SVM also gave a good result. with 55 percent sitting on his second seat and leaving behind most CNN models. Other algorithms have good results but are the lowest in succession.

All the results we get are obtained with a dataset that is not at all good. However, we have achieved very successful results. Because if someone who examines the dataset does so with his eyes, he will not get much results from the results we have obtained .

We only deal with 7 regions in Turkey. However, with the development of the project, tests on larger scales can be realized and become the first machine learning based work in the field of geographical architecture.