This is the folder that contains the code to run our CNN model.
To run the model, just run python cnn.py. It might take couple hours to run on a
personal computer, but you can significantly reduce the time by running it on a server.
test_img_generator.py is used to create new partly-blurred images to the tarpath folder.
It returns a 3D list labels which indicates the blurred parts in each image.
Our training model is saved in motionblur.h5.