Is Somfy's Facial Recognition Feature Not Working Correctly?
The facial recognition feature on Somfy security cameras is a powerful tool, designed to move beyond simple motion detection and provide smarter, more personalised alerts. It allows you to know not just that someone is at your door, but who is there. However, when this feature fails to identify a familiar face or works unreliably, it can undermine your confidence in the system.
This comprehensive guide will help you troubleshoot common problems with Somfy's facial recognition, from setup issues to improving detection accuracy, so you can get the most out of this intelligent feature.
## How Somfy Facial Recognition Works
Before troubleshooting, it's helpful to understand the basics. The system works by comparing faces it sees in real-time to a library of "known faces" that you create.
- You Create Profiles: In the Somfy Protect app, you create profiles for family members and frequent visitors, adding clear photos of their faces.
- The Camera Analyses: When the camera detects a person, it analyses their facial features.
- It Compares and Notifies: It compares these features to your library. If it finds a match, it can send you a specific notification (e.g., "Hannah has arrived home") and disarm the system if you've configured it to do so. If there's no match, it alerts you to an "Unknown Face".
## Common Problems and How to Fix Them
Let's address the most frequent issues users encounter with the facial recognition feature.
### 1. Failure to Recognise a Known Person
This is the most common complaint. If your camera is flagging your partner or children as an "Unknown Face", try these solutions:
- Improve the User Profile: A single photo is often not enough. The key to accuracy is a robust profile. Go to the "Users" section of the app and add multiple photos for the person. Include photos from different angles, in different lighting conditions, and with/without glasses if applicable.
- Check for Obstructions: Is the person's face partially covered? A hat, sunglasses, a face mask, or even hair covering the eyes can prevent a positive identification.
- Clean the Camera Lens: A smudged or dirty lens can significantly impact the camera's ability to see facial details clearly. Clean the lens gently with a soft, microfibre cloth.
- Lighting is Key: Very harsh backlighting (e.g., the sun directly behind the person) or very low light can make it difficult for the camera to capture the necessary facial details. Ensure the area is as evenly lit as possible.
### 2. Inaccurate Identifications (Mismatches)
Sometimes the system might incorrectly identify one person as another.
- Refine User Profiles: This issue also points back to the quality of your photo library. Ensure the photos for each user are distinct and clear. You can also go into the event history, find the incorrect identification, and use the feedback option (if available) to tell the system it made a mistake, which helps the algorithm learn.
### 3. Not Receiving Correct Notifications
If you're not getting alerts for known individuals, check your app settings.
- Enable Smart Notifications: Dive into the notification settings in your Somfy Protect app. Ensure that you have enabled notifications specifically for "Known Person" events. You may have them turned off, while "Unknown Face" alerts are still active.
### 4. General Performance Tips
- Optimal Camera Placement: Install your camera at a height and angle that provides a clear, head-on view of people's faces as they approach. A camera that's too high and pointing down will mostly see the tops of heads, making facial recognition almost impossible.
- Keep Firmware Updated: Somfy periodically releases firmware updates for their cameras. These updates can include improvements to the facial recognition algorithms. Ensure your camera's firmware is always up to date via the app.
By taking the time to build a good library of known faces and optimising your camera's placement and settings, you can dramatically improve the reliability and accuracy of Somfy's facial recognition feature.