Abstract
Internet of Things (IoT) has dramatically penetrated into today’s daily routine, with the number of new devices being growing exponentially, anticipating by year 2035 over one trillion of IoT devices to be connected around the world. However, IoT is still in its infant stage, regarding mainly the heterogeneity of the connected devices. In order to contribute to this gap, this paper proposes a mechanism for recognizing heterogeneous IoT devices of unknown device type, by applying specific computer vision techniques. This mechanism implements four (4) stages to successfully recognize a device of unknown nature. Initially the discovery and the connection of both known and unknown devices occur, followed by the online search of their images, so as to categorize and map the unknown devices’ images with the images of the known devices. As a result, since the device type of the known devices is a priory known, and each unknown device is mapped with one or more images of these known devices, the device type of each unknown device becomes known, as well. The prototype associated with this paper provides an example of this mechanism, demonstrating in detail each discrete stage