Microsoft is talking a lot at this year’s Build conference about bringing its artificial intelligence tools from the very edge of the network to the growing army of smart devices inhabiting our homes and offices.
One of the those AI skills is computer vision, which allows tiny cameras to recognize objects and then, armed with that knowledge, to take relevant action—for example, if a camera on a drone sees and recognizes a broken drainage pipe, it sends a fix-it ticket to the maintenance server. Microsoft and Qualcomm said today they’ll co-create a developer kit containing the hardware and software needed to build such computer vision chops into small devices.
Such devices will leverage Microsoft’s cloud-based machine learning service. Qualcomm will provide chips specially designed to support the millions of computations needed for computer vision tasks.
Separately, Microsoft announced that its Custom Vision computer vision service can now run on devices at the edge of the network (such as on IoT devices with cameras), allowing them to recognize objects and take actions without having to connect with the cloud.