Object detection locates instances of objects that belong to pre-defined classes, such as "car" or "van". This is more general than object recognition, which recognizes specific objects such as a red van that matches a specific model and has the logo of "ABC company" painted on its side.

Object detection differs from image classification because it returns the position for every detected object. Object detection can locate multiple objects within the analyzed image or region, whereas classification returns a single classification result and no position information. For example, using an object detector that has been trained to locate cars and people might return more than one instance of a car and more than one instance of a person in the same image.