Don't leave your Someones as Someones! 📸
Friendly Faces is trained as it records Event videos, detects faces (Someones), and you name those faces. It won't learn if it doesn't have a growing database of named faces, so keep them updated.
💡 Keep the feature enabled so your camera(s) can detect and learn new faces.
To train the Friendly Faces feature:
When training it to recognize faces, we recommend being within 2 meters of the camera and directly facing the lens. Attempting various distances and face orientations will also enrich the database.
Once your camera(s) start populating faces in the app, you must tag the images with the individuals name - don't leave them as Someones. To do this, go to Account > Friendly Faces.
All faces captured will be displayed on that screen. Tap on the image to:
- Edit the image (add a new or existing name)
- Select and remove faces
- Search through the last 14 days to identify and review face-detected events.
I just got Friendly Faces and it's not naming everyone properly. Now what?
It's learning! This feature takes training and time to build the database and attach the right names to the right faces. Go to Account > Friendly Faces and edit, name, or remove faces. The more faces you add and name, and the higher the quality images, the better the AI will get at identifying the people.
Additional Notes:
- The app will continue to collect new faces as long as people are detected. Assign the images new or existing names, or remove the ones you don't want.
- The higher the quality of images captured, the more accurate the algorithm will be.
- Friendly Faces takes time to start recognizing faces consistently and accurately.
- If a person's face is in the frame clearly, for long enough, and is stored in the database, the AI will know who the person is!
- 💡 There are two ways this is reflected: notifications and filtering on faces in the Events tab.