Getting the Most Out of Ezviz Facial Recognition
Ezviz cameras with facial recognition offer a powerful way to make your security alerts smarter and more meaningful. Instead of getting a generic "motion detected" alert for every passing car or swaying tree branch, these intelligent cameras can specifically notify you when a person is seen, and some can even identify who that person is.
However, when this feature doesn't work as expected, it can be confusing. This troubleshooting guide will help you understand how facial recognition works, how to optimise its performance, and what to do when it's not behaving correctly.
Understanding Ezviz's AI Detection Features
It's important to know the difference between the two main types of smart detection on Ezviz cameras:
- Person Detection: This is a common AI feature that identifies a human shape within the camera's view. Its goal is to differentiate people from other moving objects like pets, vehicles, or shadows, thereby reducing false alarms.
- Facial Recognition: This is a more advanced feature. After first detecting a person, the camera's processor analyses the facial features. It can then compare this face against a gallery of known individuals that you have created in the app, allowing for personalised alerts (e.g., "Family member has arrived").
Why Facial Recognition Might Not Be Working
If you're getting inaccurate results or the feature isn't triggering at all, consider these common causes.
1. Sub-Optimal Camera Placement and Lighting
This is the most frequent cause of poor performance. The camera needs a clear, well-lit view of a person's face to analyse it properly.
- Camera Height and Angle: Mount the camera so it's looking at people head-on, not from a steep top-down angle. A height of around 2-3 metres is often ideal.
- Lighting Conditions: Strong backlighting (e.g., pointing towards the sun) can create silhouettes that make faces impossible to see. Similarly, deep shadows or very low light at night will prevent the camera from capturing the necessary detail. Ensure the target area is evenly lit.
- Distance: Faces that are too far away will be too small for the camera to resolve the details needed for recognition.
2. Configuration and Settings Issues
The feature may simply be disabled or misconfigured in the Ezviz app.
- Enable the Feature: Dive into your camera's settings in the app and ensure that "Person Detection" or the more specific "Facial Recognition" mode is turned on.
- Detection Area: Make sure you haven't drawn a motion detection zone that accidentally excludes the area where people's faces would appear.
- Firmware Updates: Check that your camera's firmware is up to date. Ezviz often releases updates that improve the performance and accuracy of their AI algorithms.
3. Obstructions and Face Coverings
The algorithm needs a clear view of the entire face.
- Physical Obstructions: Plants, decorations, or other objects can block the camera's view at the critical moment a person walks by.
- Face Coverings: Hats, sunglasses, and face masks can obscure key facial features, making it difficult or impossible for the algorithm to make a positive identification.
Steps to Improve Accuracy
- Review Camera Placement: Watch your camera's live feed and imagine you are the AI. Can you clearly see the faces of people walking by? If not, adjust the camera's position or angle.
- Optimise Lighting: Consider adding external lighting, such as a porch light or motion-activated floodlight, to improve performance in the evening and at night.
- Build Your Face Gallery: For facial recognition models, take the time to build a good library of known faces. Add clear, well-lit photos of family members from multiple angles to give the AI the best possible chance of making a match.
- Keep the Lens Clean: A smudged or dirty camera lens can significantly degrade image quality. Wipe it clean with a soft microfibre cloth regularly.