Page 320 - AI Computer 10

u Convolutional Layer

The first layer of a CNN which is used to extract features of an image with the help of convolution operations.
u Feature Map
The output matrix produced after a series of convolutions on an input image matrix.
u Rectified Linear Unit Function (ReLu)
A non-linear activation function that is commonly used in CNNs to remove all the negative numbers that
exist in a feature map.
u Pooling Layer

The layer in a CNN responsible for reducing the spatial size of the convolved feature map while still retaining
the important features.
u Fully Connected Layer
The last layer in a CNN that performs the task of classifying the images on the basis of input.

I In a n a NNutshellutshell

• Computer vision can be defi ned as the domain of AI that enables computers to analyse meaningful
informati on from images, videos, and other visual inputs.

• Image resoluti on, which expresses the amount of informati on in digital pictures, is measured in pixel per
inch (PPI) or dot per inch (DPI).

• Grayscale image is an image representati on mode in which the colour of each pixel has a range of shades of
gray without apparent colour.

• Three diff erent channels called the R channel, G channel, and the B channel are used to store an image in
RGB mode in a computer.

• The term “RGB” stands for Red, Green, and Blue colour and it refers to three hues of light that can be mixed
together to create diff erent colours.

• OpenCV, an open source tool developed by Intel, has a cross-platf orm library through which you can develop
real-ti me computer vision applicati ons.

• neural networks refer to the systems of neurons which are either organic or arti fi cial in nature.
• Convoluti on Neural Network is a neural network which serves as a powerful technology in many computer
vision applicati ons.
• The primary diff erence between CNN and other neural networks is that CNN takes input as a two-dimensional
array and operates directly on the images rather than focusing on feature extracti on.
• A Feature helps in reducing the image size to make processing faster and effi cient, and focusing on several
features that can help in processing the image.
• ReLU is a non-linear acti vati on functi on that is commonly used in deep neural networks to remove all the
negati ve numbers that exist in a feature map.
• The Pooling layer in a CNN makes the image smaller and manageable, and reduces the chances for small
transformati ons, distorti ons, and transiti ons.
• No-code AI tools for computer vision enables individuals and organisati ons to develop and implement image
and video analysis soluti ons without requiring extensive programming skills.
• No-code AI tools employ user-friendly interfaces, featuring drag-and-drop functi onality that empower users
to build, train, and deploy computer vision models effi ciently.

186
186