Google Home update uses clothing and sound to improve camera alerts

In short: Google Home is getting an update that aims to reduce wrong camera alerts by using cues like clothing and by describing important sounds.

What happened

Google says a new Google Home update, starting June 23, will make its “Familiar Faces” feature better at identifying people in your home. Familiar Faces is the tool that lets you tag household members so your camera can tell who is who.

Until now, the system mainly relied on faces. With the update, Google says it can still recognize tagged people even when their face is not clearly visible, for example if they are turned away. It does this using “non-biometric signals,” like body size and clothing color (similar to how you might recognize a family member from behind by their jacket).

Google is also changing how the Familiar Faces library is maintained. It will automatically update with newer images of the people in your home, which should help prevent mistakes caused by old or unhelpful reference photos.

In addition, Google says its AI-written video event descriptions can now identify specific sounds, such as dogs barking, alarms, or footsteps. Those sounds can be added to the event notes even if the camera does not capture what caused the noise.

Why it matters

For many people, security camera alerts are only useful if they are accurate. If your phone keeps warning you about the wrong person, or describing events that did not happen, it becomes easier to ignore alerts. Google’s changes are meant to make notifications more reliable and more descriptive without you having to constantly retag people or check every clip.

Source: The Verge AI