This is a simple Deep neural network trained to solve a binary classification problem, which is deciding whether some picture represent the city of Algiers or not.
The data used was gathered using Download All Images Extention The result can be found in /data folder.
The model used has 3 blocks of convolutional layers and Max pooling layers, The two final layers are a flatten layer and a dense one. The last layer has one unit whose objective function is Sigmoid (between 0 and 1).
The resulting plots can be found on the code file.
When we feed the model with two pictures (one of Algiers and one of not) that can be found in /test. The result is like so:
The above picture is a picture of Algiers, we can see that the model has correctly predicted it
The above picture is not a picture of Algiers, but rather of Constantine. We can see that the model has correctly predicted it
