To train the computers learn a certain types of patterns, different types of dimensions or shapes and attributes, images are used with proper annotation that can be done with the help of a software to outline the entire object making it visible to machines or computer vision.
