Python version used in this project is: Python 3.5
Modules and libraries used in this project are:
Keras - to implement our CNN opencv - for preprocessing numpy - for some preprocessing and neccessary operations
You can view the jupyter notebook i included for a step by step implementation, and of course you can still use the files there.
Run the Main.py file and pass it an arugment which is a picture and you will get the score of the possibiltiy of having a car in the picture you passed, you can use the pictures in the testImages folder.
The datasets i used are not included in the folder because the size was way too big, however i used two datasets which can be found here
First dataset for car images:
http://ai.stanford.edu/~jkrause/cars/car_dataset.html
Second dataset for non-car images:
http://www.ais.uni-bonn.de/download/datasets.html
the second dataset for noncars images have some car pictures, since i used this dataset as examples of what a car does not look like for my model i excluded every picture that has a car in it
Also please note that i didn't use the entire two datasets, i used around 4500 pictures from each dataset, which results in around 8900 pictures from both datasets.