In the last months there has been a big excitement about Watson visual recognition.
I visited many customer site to explain and to create few proof of concept to address many traditional scenario in a very innovative way. I’m applying the Watson visual recognition to:
I am currently exploring how to check the external condition of a van/truck by visual recognition. For example when a fleet vehicle comes back into company’s hub, by a camera system automatically it is possible check if the van has a damage to the bodywork such as scratches, dents, or other things of this kind.
Of course in this simple proof of concept I used only two classes, that of the damaged vehicle and that of the perfect vehicle. In the samples of the “damage class” I used only dents and scratches.
I built the PoC by simple nodered flow, a dashboard based on freeboard and naturally by Watson visual recognition api.
Here is a demo from my Bluemix environment: