Notes : 9.3A Visualizing CNN layers' Filter Patterns

1. Create a model by API function and sequential model
2. model fit_generator
3. Plotting our loss charts
4. Plotting our accuracy charts
5. Save model
6. keras.preprocessing.image.load_image which returns PIL image
7. Convert image to numpy array
8. Display a two dimensional array as image
9. Creating a new model from an input and output tensors of existing model
10. model predict which returns the output of each layer.
11. For each Conv layers' output, visualizing every 2D numpy array to observer conv filters' patterns