How to Fix Inaccurate Doorbird Person Detection
Receiving a notification that a person is at your door only to find it was a swaying tree, a passing car, or just a shadow can be very frustrating. While Doorbird's motion detection is a powerful tool, it can sometimes be triggered incorrectly. This guide will help you understand why this happens and what steps you can take to fine-tune your system for more accurate person detection alerts.
By making a few adjustments to your settings and considering your device's environment, you can significantly reduce the number of false notifications.
Common Symptoms of Incorrect Person Detection
If your Doorbird is struggling with accuracy, you're likely experiencing one of these issues:
- Alerts for Moving Shadows: You receive notifications on windy or sunny days caused by the shadows of trees, flags, or other objects.
- Notifications from Car Headlights: Passing cars at night trigger person alerts as their headlights sweep across your property.
- Detections of Animals or Other Objects: Large pets, falling leaves, or even heavy rain are sometimes flagged as a person.
- No Detection of Actual People: In some cases, the system may fail to alert you when a person actually approaches your door.
- Inconsistent Performance: The accuracy seems to vary wildly depending on the time of day or weather conditions.
Step-by-Step Guide to Improve Person Detection Accuracy
Follow these steps to systematically reduce false alerts and improve the reliability of your Doorbird's person detection.
1. Adjust the Motion Sensor Sensitivity
The sensitivity setting determines how much thermal and visual change is needed to trigger an event. When it's too high, it can be triggered by insignificant events.
- Access the Settings: Open the Doorbird app and go to Settings → Motion Sensor.
- Decrease Sensitivity: If you are getting too many false alerts, lower the sensitivity level by one step.
- Test the Change: Allow the system to run for a few hours or a day to see if the frequency of false alerts has decreased. You may need to adjust it a couple of times to find the perfect balance between responsiveness and accuracy.
2. Refine Your Motion Detection Zones
This is one of the most effective tools for eliminating false alerts from predictable sources like a busy street or a neighbour's property.
- Define a Precise Area: In the Motion Sensor settings, you can draw one or more zones where motion should be detected. Be very specific.
- Focus on a 'Choke Point': Draw a tight zone directly around your doorstep, walkway, or porch area where a person must walk to approach your door.
- Exclude Problem Areas: Actively exclude public pavements, roads, and any trees or bushes that move a lot in the wind. This tells the system to ignore any motion that occurs in those areas.
3. Check the Physical Environment and Mounting
The camera's position and condition can have a major impact on its performance.
- Clean the Lens and Sensor: A smudged camera lens or a dirty PIR sensor dome can distort the view and affect thermal readings. Regularly clean both with a soft, microfibre cloth.
- Avoid Direct Sunlight: If possible, ensure the camera is not pointing directly at the rising or setting sun. Intense sunlight and lens flare can confuse the video analysis.
- Check for Heat Sources: The PIR sensor detects heat. Ensure it is not aimed at an air conditioning exhaust vent, a dryer vent, or a patch of dark ground that gets very hot in the sun, as these can cause false triggers.
4. Keep Your Firmware Updated
Doorbird continuously works to improve the algorithms that power their motion and person detection.
- Check for Updates: In the Doorbird app, navigate to Settings → Administration → Firmware Update.
- Install the Latest Version: If an update is available, install it. This is a simple step that can often lead to a noticeable improvement in detection accuracy without any other changes.
By carefully tuning these settings, you can train your Doorbird to be a much more reliable observer, ensuring you are only alerted to the events that truly matter.