How can a machine be able to see and what does this entail? Computer Vision is a branch of Artificial Intelligence that has been exploited in recent years thanks to advances in the field. The first camera was invented over 100 years ago and since then the components have improved exponentially so that today we have incredible resolutions at long distances. So this is the first step to enable Computer Vision, a camera that captures images so they can be processed and analyzed by a computer. It is natural for humans to see a picture of a mountain and easily distinguish it from a tree, but how can computers possibly differentiate them?
A machine perceives an image as a huge set of integer numbers that represent the intensity of the colors throughout the picture. Using these numbers and Machine Learning, a computer can compare the patterns in the image to understand what all those numbers mean in a specific order.
Since we use machine learning to train the algorithm, this means that the more information you have to compare and analyze, the better accuracy you can get when recognizing objects. These groups of information given to the machine is what we call a Data Set, which is then processed by a Neural Network that predicts the output and determine what is in the image, there is alway an error margin that can be reduced with more Data Sets and therefore, more algorithm training.