Google is set to roll out an update to its Google Home system, aimed at improving the accuracy of its facial recognition feature. As of June 23rd, the system will utilize additional non-biometric signals, such as body size and clothing color, to identify individuals in the Familiar Faces library, even when their faces are not clearly visible. This enhancement will reduce instances of mistaken identity, particularly when users are facing away from the camera. The update also introduces automatic updates to the Familiar Faces library, ensuring that the most recent images of household members are used, thereby minimizing inaccurate notifications. Furthermore, the system's AI-generated video event descriptions will now be able to identify specific sounds, such as dogs barking or alarms, and include them in the event logs, even if the audio source is off-camera. These updates address issues previously encountered with the system, including event logs with incorrect descriptions and non-occurring events. Additionally, the latest Google Home app update, version 4.20, includes new System Health alerts for Nest thermostat issues and improved support for Matter switches, demonstrating Google's ongoing efforts to refine its smart home ecosystem.

Source: The Verge