See all Blogs

Shahroz Bakht

Computer Vision – What it is?

What Is Computer Vision?

A field in artificial intelligence and machine learning that deals with the tools and technologies that provides a way for computers to visualize and gather visual information from the real world is known as Computer Vision.

All one requires to see the world is a camera but simply connecting the camera and receiving pictures is not enough. The real challenge is how you would classify and interpret objects that are occurring in images and videos, how to identify the relationships between them and understanding the context of what is going on. Computer Vision deals with computers being able to explain what is in an image, video footage, or real-time video stream.

How Does Computer Vision Work?

The machine needs to be trained on thousands of examples in order for the machine to recognize visual objects. For example, if you want a person to distinguish between two objects, namely a car and a bicycle, how would you describe this to a human?

A normal statement would be that a bicycle contains two wheels whereas a car has four wheels or a difference can be that a bicycle has pedals and the machine does not have them.

Lighting Difference

An important process for computer vision is to gather all knowledge about the real world that distinguishes between objects in terms of lighting. A filter might make a ball look blue or yellow while in fact it is still white. A red object under a red lamp becomes almost invisible.

Noise

A hard trait in object detection for computer vision is when there is a lot of noise factor present. Noise can be regarded as pixels in the image when they appear brighter or even darker than they normally should be. For example, videocams that detect violations on the road are much less effective when it is raining or snowing outside.

Different angles

It is paramount to have several angles of a picture or else the computer will not be able to recognize when and if the angle within a picture changes.

Overlapping

Overlapping occurs when there is more than one object on a particular image. This way some of the traits of the objects may stay hidden and this provides added difficulty for the machine to recognize objects.

Various Kinds of Objects

It is very common that objects that belong to the same category may possibly look different. For example, there are many categories of lamps but the algorithm embedded must successfully know the difference between a nightstand lamp and a ceiling lamp.

Faux Similarity

Sometimes different items from different categories may look entirely similar. For example, somedays you might have met people that remind you of a celebrity if there photos are taken from certain angles but in real life they look entirely different. In another example, samoyed puppies can be easily mistaken for little polar bears in some pictures, cases of misrecognition are common.

Utilizes of Computer Vision

Translating digital images and videos comes in neat in many fields. Let us look at some of the use cases:

  • Medical diagnosis.
  • Factory management.
  • Retail.
  • Security systems.
  • Animal conservation.
  • Self-driving vehicles.

Summing Up

An innovative field that utilizes the latest machine learning technologies to construct software systems whose motive is to assist humans across different fields is Computer Vision. This ranges from retail to wildlife and smart algorithms constructed can solve the problems that exist in image classification and pattern recognition, it can supersede human effort as well.

  See all Blogs