Lattice Blog


Proven Human Presence Simplified

Person Detection at sub 1 mWatt
Posted 01/11/2019 by Hussein Osman

Posted in

With the evolution of SMART cities, homes and factories, there is a multitude of situations that demand accurate proof of the presence of a person…and perhaps his or her positioning and actions. In a modern factory using Industrial IoT (IIoT)/ Industry 4.0 practices, safety must not be left behind. Is there a person present within the “danger zone” of a large machine? Or maybe we just need to know whether someone is standing in front of an ATM machine so that a screen is activated from a power-saving sleep mode.

With new, highly-capable, edge AI devices that do not require connectivity to the cloud, it is becoming simple to embed neural network-based image detection systems that deliver differentiated, marketable features. For example, with the Lattice sensAI solution stack, designers can enable an image sensor-based application to detect and confirm any item – face, animal, object – in real-time, leveraging an advanced neural network engine. So, no longer must we rely on powerful cloud computing servers requiring constant internet connectivity and server access costs to enable this functionality.

Now, we can tap the parallel processing capability of FPGAs to bring more computational resources to the network edge without significant architectural changes to existing devices. Designers are using sensAI hardware and software tools to build high performance AI devices without violating power, footprint and cost constraints. With Lattice sensAI reference designs and demos for common AI use cases like object detection, object counting, human presence, and key-phrase detection, we can move fast to build custom solutions in rapidly shrinking time-to-market windows.

Learn more