Computer vision

What is computer vision?

We the humans have eyes, we can capture our surroundings, process it and understand it. Using computer vision we can train the machine to do the same thing. Computer vision is a field of artificial intelligence which deals with training the machine to interpret and understand the visual world.

How does computer vision works?

Machine can only understand the numbers. It identifies the images by the number patterns in an array. 

So only one image in a specific angle or color is not enough to train the model. For example if you are training the model to detect cars with a single image of a car, then your model can't detect the cars which is in different angle, color, surroundings etc. So you just have to collect multiple images of cars and you have to annotate it as a car then convert that images and annotations into binary records based on the library you are using. So that it can just focuses only on the image inside the bounding boxes and whenever a new image comes in then the model will detect it as a car.

The most popular algorithm used is CNN(Convolution Neural Network) which works by breaking an image into smaller groups of filters and each filter is a matrix of pixels. Now it will do series of calculation on pixels to find the pattern. CNN contains many layers, the first layer of CNN is able to detect high level patterns like rough edges and curves, likewise it goes on and at the end it can identify the image. It needs large amount of labelled training data so that it can randomize the filter value. Each time CNN makes prediction on labelled data it uses the loss function to compare how close its prediction was to the image actual labelled and update the filter values.

We can also train the model with videos where RNN(Recurrent Neural Network) comes into play.

Applications of computer vision:

1) Medical image analysis.

2) Automobile industry.

3) Various object detections and recognition.

4) Security purposes - facial recognition and all.

5) Multimedia.

6) Navigation etc.

Opencv:

Opencv is a computer vision library with which we can do real time computer vision. This library supports python, C++, java etc. Trained machine learning models can also be deployed using this library that is it helps to detect objects or faces, to classify human actions and so on by reading the images or videos from the specified path or from the real time. Also these images and videos can be processed by some techniques like image stitching, contours, edge detection etc. 3D graphs and models can be extract using this library.

About srinila's view

we are here to explain about the booming technical field Artificial intellingence and the fields such as data science, machine learning, deep learning etc. which concerns with AI

Related Post