object identification

Terms from Artificial Intelligence: humans at the heart of algorithms

Page numbers are for draft copy at present; they will be replaced with correct numbers when final book is formatted. Chapter numbers are correct and will not change now.

Object identification follows object recognition in image processing and is the process of working out what an object is. This may be at the level of kinds of things: dog, cat, car; or more precisely identifying a specific car or person. This may use some form of template matching against known images. Of course, the way an object appears depends on how far away you are, maybe the zoom setting of the camera, and if the object is facing you, side on, etc. Sometimes this is dealt with by normalisation of the object orientation and size. For neural networks the same initial training image may be transformed to different orientatiosn and scales to grow the training data during the training phase in improve object identification.

Used in Chap. 12: page 191