**Convolution Neural Network **or CNN for short is deep learning model which learns pattern in the images. It is widely used for image classification tasks.
What is image classification: Image classification is process of assigning labels to images according to their types (classes).
Consider we have images of digits from 0 to 9:
So when we provide image of given class to image classification model (in our case CNN), the model will assign label to images according to the class.
For example if we provide the image given below to the our model.The model should assign the label “6” to the image.
CNN is divided into two parts; feature learning and feature classification.
Feature learning:
In feature learning part, starting layers in the model learn the simplest features in the images like edges, corners etc; and subsequent layers learns little bit complex features like parts of object; and the layers after them learn more complex features like shape of object. These layers are called feature extraction layers.
Feature classification:
In** feature classification** part model learn the function which is used to classify images. Fully connected layers in this part select the best features from the feature learning part and apply some function (e.g soft max) on final feature vector to calculate the probability of image as being digit between 0 to 9.
#convolution-neural-net #deep-learning #classification #artificial-intelligence #tensorflow
