An intuitive guide to CNN. Introduction and working process

1. Convolutional
Neural
Network
• Abhishek Sharma (1551244)
• Sagar Sangam (1551207)
• Deepak Prasad (1551060)

2. What is CNN?
Why CNN?

3. What do we
have?
• Computer vision
• CNN into picture and it’s history
• Architecture and Working

5. Human Vision

6. • The vision begins with eyes, but truly take places in the
Brain
• It took mother nature over 500 million years to create
a system to bring the ability of recognizing things.
• The collaboration between the eyes and the brain,
called the primary visual pathway, is the reason we
can make sense of the world around us.

7. How does the computer see the world?

8. Human Vision Computer Vision

9. • Computer ‘see’ in a different way than we do.
• Every image can be represented as a 2-dimensional
arrays of numbers, called pixels.
Pixels

10. • What to do with these pixels?
• Ultimately we want machines to see like we do naming
objects identifying peoples, inferring 3d understanding
emotions, actions and so on.
• The first step towards this goal is to teach the
computer to identify objects and for this the only pixels
are not enough.

11. Teach a computer to identify a Cat

16. • Even something a simple as a household pet can present an infinite
number of variations to the object model, and that ‘s just one thing.
• Similarly to how a child learns to recognise objects, we need to show an
algorithm millions of picture before it is able to generalize the input and
make predictions for images it has never seen before.
• Here comes Convolution Neural Network into picture.

17. What is CNN?

18. • To teach an algorithm how to recognize objects in images, we use a specific
type of Artificial Neural Network: a Convolutional Neural Network (CNN).
• It is a special type of Neural Network widely used for recognize the image.

19. Where it all begins ?
• Convolutional Neural Networks are inspired by the brain. Research in the 1950s and
1960s by D.H Hubel and T.N Wiesel on the brain of mammals suggested a new model
for how mammals perceive the world visually.
• In 1980, a researcher called Fukushima proposed a hierarchical neural network model.
He called it the Neocognitron.
• The Neocognitron was able to recognise patterns by learning about the shapes of
objects

20.  The ‘pixels’ of the object used as the input to CNN to recognize the object; earlier the
only pixels were not enough to identify a certain object.

21. Why is CNN?

22. Regular Neural Network and CNN

23. Architecture
Convolutional Neural Networks are a bit different from Regular Neural Network.
First of all, the layers are organized in 3 dimensions: width, height and depth.
Further, the neurons in one layer do not connect to all the neurons in the next layer
but only to a small region of it.
Lastly, the final output will be reduced to a single vector of probability scores,
organized along the depth dimension.

24. CNNs have two components:
• The Hidden layers/Feature extraction part
In this part, the network will perform a series
of convolutions with RELU and pooling operations during which
the features are detected.
• The Classification part
The fully connected layers will serve as a classifier on top of these
extracted features. They will assign a probability for the object on
the image being what the algorithm predicts it is.