Object detection systems combine the tasks of image classification (identifying the type of an object in an image) and object localisation (finding the location of one or more objects in an image).

An object detection model can be trained to detect the presence and location of multiple classes of object. However, a particular model will not be able to recognise objects that it has not been trained to detect. For example, a model trained only to detect people will not recognise cars or animals.

Each cloud service uses their own model trained to recognise specific objects, so results may vary.

All images analysed using this demo are stored for 24 hours and then automatically deleted using Microsoft's Azure data lifecycle management.

Object detection systems have been applied in a wide range of fields: the current generation of prototype self-driving cars make extensive use of the ability to identify cars, pedestrians and road signs surrounding the car; ball tracking and player tracking is prevalent in many sports, and used for in-game decision making and post-game analysis; and the robotics industry relies on robust object detection systems to allow robots to react quickly and accurately to chances in the environment.

Within an educational setting, object detection could be used for tasks such as automatically assessing room occupancy, to optimise the use of classrooms and lecture theatres, or estimating lecture attendance numbers.

As with much technology of this kind, object detection systems depend almost entirely on the types of images used to train the models. Any biases or omissions in the images used to train a model will be reflected in the performance of the object detection system, and may result in, for example, the system being completely unable to recognise particular kinds of objects in an image.