How to use the AI face detector
- Click "Choose an image" and select a photo from your device.
- Click "Detect Faces". The first use downloads the model, which may take a few seconds.
- Read the face count and view the boxes drawn around each detected face.
- Click "Download annotated image" to save a copy with the boxes marked.
How face detection works
This tool runs MediaPipe's face detection model, a compact neural network that executes entirely in your browser using WebAssembly. When you select a photo, it is drawn to a canvas and passed to the model. The model returns a set of detections, each with a bounding box that describes where a face sits in the image. Those coordinates are normalized between zero and one, so the tool multiplies them by the image width and height to draw accurate rectangles at full resolution.
What the results mean
The face count tells you how many faces the model found, and each box marks the approximate position and size of a face. The model is optimised for speed rather than recognition, so it is quick but does not identify anyone. It simply reports that a face is present and where it is — useful for checking group photos, cropping around people, or counting heads in a crowd.
Common uses
Photographers use face detection to confirm everyone is in frame before a shoot. Social media managers use it to check that thumbnails centre on a face. Developers test it as a building block for larger projects, and anyone can use it to quickly count the people in a group photo without counting by hand.
Privacy
Everything runs locally in your browser. Your photo is never uploaded, transmitted or stored on any server, and the model does not recognise or name anyone. The detection happens on your own device, which keeps your images private.