Have 2 g5 cameras installed on the front of my house. I keep getting false package detection alerts. I have lowered the sensitivity but the camera is still struggling. The issue is that my steps are older bricks that I guess are triggering something in the logic. It would be great if there was some process of capturing the steady state of the steps so that it can better detect changes
@kevin It sounds like you’re having a bit of a challenge with your G5 cameras and the false package detection alerts. Your idea of capturing the steady state of your steps to better detect changes is quite insightful. From the information available, it seems there isn’t a specific feature that captures the steady state for package detection on the G5 Pro cameras. However, you might want to try a few adjustments:
-
Re-evaluate Installation Position: Ensure the cameras are positioned at an optimal height and angle to minimize interference from the steps. Sometimes, the angle can cause items like bricks to be misinterpreted as packages.
-
Adjust Sensitivity Settings: While you’ve already tried lowering the sensitivity, sometimes finding a balance between sensitivity and accuracy requires a bit of experimentation.
-
Automation and AI Features: Consider exploring any AI features or automation conditions that might help you better manage detections. For instance, utilizing features like “Recognize the specified face” or “Recognize stranger’s face” might help in reducing false alerts if they are being triggered by human-like shapes.
While these suggestions might not perfectly solve the issue, they could help in reducing false triggers. If the problem persists, I’d recommend reaching out to the community for additional tips or contacting Aqara’s customer service for more targeted support. Keep experimenting, and feel free to share any solutions you find!
1 Like
@kevin Sorry to hear you’re dealing with false package detection alerts on your G5 cameras—those older brick steps can definitely throw off the AI logic! Let’s try a few troubleshooting steps to improve this.
First, ensure the camera is installed at a height of 3 meters or below, with the lens tilted downwards at a 30° angle—this helps the AI better distinguish between static objects (like steps) and actual packages . Also, make sure your camera’s firmware is updated to the latest version, as updates often refine detection algorithms . If the issue happens at night, turning on the spotlight for extra lighting might help the camera clarify details .
You mentioned lowering sensitivity, but for AI detection, the recommended setting is actually the highest sensitivity to improve recognition—you might want to try adjusting it back and see if that helps .
As for capturing a “steady state” of the steps to reduce false triggers, that’s a great suggestion! Currently, this feature isn’t mentioned in our available resources, but if you’d like to share this feedback, you can submit it through the Aqara Home App (Profile>Help&Feedback>Customer Service) for our team to review.
Hope these tips help—feel free to share how it goes, or if you try other adjustments that work! 
@kevin I’m sorry to hear you’re dealing with persistent false package detection alerts on your G5 cameras. While the reference information doesn’t specifically address capturing a “steady state” for static objects like brick steps, here are a few general tips from Aqara camera troubleshooting that might help: ensuring the camera is installed at a recommended height (3 meters or below with a 30° downward tilt) and updating the firmware to the latest version could potentially improve detection accuracy .
Since you’ve already adjusted the sensitivity, if the issue persists, consider reaching out to the community for shared experiences—other users might have found workarounds. If you have suggestions for feature improvements like capturing a steady state, feel free to submit feedback through the Aqara Home App (Profile>Help&Feedback>Customer Service) for our team to review.
Thanks for sharing your experience, and I hope we can find a solution that works for you soon!