Computer Vision is a segment of artificial intelligence that is concerned with the recording and interpretation of video by a program to allow a system to do something with that data like respond to something in the video feed. Some examples would be a self driving car that responds to driving conditions while its video feed shows it where other cars, obstacles, and pedestrians are located. Another example would be a robot that uses its vision interpretation to move around and to grab items with its robotic arm. Yet another example would be a farming/gardening drone or implement that moves around from the bottom x coordinate and the bottom y coordinate to the top x and y coordinates of the gardening container. It moves around and finds where the stems of plants make their way out of the earth and deposits water and possibly minerals there for the plant to consume.
The programmatic asset that is used in identifying features and objects in computer vision comes from trained deep learning models. Deep learning involves training a convolutional neural network on a set of images or video that contains an indentifiable object with features that identify it. The model learns features of the objects to identify more objects in its use. The model is stored locally on the computer with the code and the code calls the model with the computer vision data to classify whether and where the object is located in the data.
Cameras are used in computer vision that view the surroundings of a vehicle or something that moves around or possibly does not. The cameras take photographs or usually take video of the surroundings or what is in front of the drone, robot, or vehicle that is taking in the video feed. Cars have lidar and radar to make double and triple sure that the object is detected by using similar techniques and methods as with the video feed to make and match detections of the position of obstacles that are around.