DEV Community

Cover image for CNN in Computer Vision
He Codes IT
He Codes IT

Posted on

CNN in Computer Vision

CNN in Computer Vision stands for the convolutional neural network. Convolutional Neural Network is a concept emerged from human Neuron. Human Neuron consists of Dendrites, Cell body, Axon and synaptic Terminal. Dendrites are basically input, Cell body is the thinking machine, axons transfer the input and synaptic terminal is basically the output.

Is an algorithm for supervised learning of binary classifiers. Perceptron rule : automatically learn optimal weight coefficients. Perceptron recieves multiple input signals sum them up and if the sum is greater than threshold it returns output, else it don’t return any signal.

Convolutional Neural Network
CNN is used in computer vision for the purpose of what values should be used in a matrix so it provide a best filter. For example if we have to identify edges in an image where there are only vertical edges present, and we are providing horizental edge detection kernel, then it is of no use. While Convolutional neural network learn from the training set and apply on the test set what best fits on it.

CNN in Computer vision has Input Layer consists of matrix of an image. Weights connect to input layer and Output layer connects to Weights.

There are many values we get after we do convolution. For example if we have 9×9 image and we do convolution on it using 3 filters, We will get three 7×7 images. Which are 147 values. To deal with this we Use Max Pooling.

To read more visit
Image description

Discussion (0)