How the age estimate works
The page loads two small neural networks into your browser from this site: a face detector and an age-and-gender network from the open-source face-api project (trained on public face datasets). We use only the age output. The network looks at texture and shape cues such as skin, eye area and facial proportions, and predicts the age it associates with similar faces. It is not reading your birth date, and it has no access to anything except the photo you pick.
Why apparent age is not real age
Age models are trained on photos labeled with real ages, so they learn what faces of each age tend to look like on average. Any face can look older or younger than average. Expect differences of several years in either direction. Lighting is the biggest factor: harsh overhead light deepens lines and shadows, while soft front light smooths them. Makeup, facial hair, glasses, expression, image sharpness and camera distance all move the estimate. Models like this are also known to be less accurate for some age groups and skin tones, depending on the data they were trained on.
How to get a steadier estimate
- Use a sharp photo with soft, even light on your face.
- Face the camera with a relaxed, neutral expression.
- Keep glasses off and hair away from your face.
- Try two or three photos and look at the middle value.
What makes a face look younger or older?
Studies of perceived age point to skin texture and evenness, facial volume, and contrast around the eyes and lips. Sun protection, sleep and good lighting in photos affect how old you look far more than any number on a screen. For proportions rather than age, try the attractiveness test or the face shape detector.
Privacy
Your photo is processed by a model running in this browser tab. It is never uploaded, and we never see it. The age model files come from this website, not a third-party service. See the privacy and biometric notice.