Feature Request: Person Detection Zones for Camera Hub G3 (Exclude Static Objects like Mannequins)

Hello Aqara Team,

I’d like to submit a feature request for the Camera Hub G3 related to AI Person Detection.

Current Behavior

  • Motion detection supports configurable detection zones, which works well.
  • Person detection does not support zones and always analyzes the full camera frame.
  • As a result, static human-shaped objects (in my case, retail mannequins) are repeatedly detected as people.
  • Because there is no way to exclude parts of the image, person detection must be disabled entirely to avoid false positives.

Use Case

In a retail environment, mannequins are permanent fixtures and cannot realistically be removed or repositioned. While motion detection zones help, they do not apply to AI person detection, which makes the feature unusable in this scenario.

This limits the effectiveness of person detection for:

  • Retail stores
  • Showrooms
  • Museums
  • Any space with static human-like objects

Feature Request

Please consider adding Person Detection Zones (or AI exclusion zones), similar to motion detection zones, allowing users to:

  • Define areas of the image that should not be analyzed by person detection
  • Exclude static objects such as mannequins, statues, posters, or displays
  • Keep person detection enabled for the rest of the scene
5 Likes

@Outburstin You seem to be experiencing an issue with AI person detection false positives due to static human-shaped objects like mannequins. For feature suggestions and improvement ideas, you can submit them through customer service. Contact the Aqara Home App Customer Service Team (Profile> Help & Feedback > Customer Service) or scan the QR code:


This will be more efficient. If you do not have an Aqara Home account, “Please click here” to create a ticket.

This should have been a feature that needs to be updated immediately. Thank you for reminding my friend. In addition to supporting regional selection, we have other logic optimization solutions. That is, after recognizing a stationary target, it will not trigger unless it is moved again. This can also reduce most ineffective event reporting.

2 Likes