Skip to main content

Every face

Group Photo Age Detector

Updated 4 min read

Drop in a group shot, a family photo or a class picture and the detector finds every face it can, numbers them largest first, and gives each one its own age estimate, age band and male or female read with a confidence score. A summary above the list shows how many faces were found and the average, youngest and oldest estimate, so you can settle the "who looks oldest" argument in one go.

Use photos of people who are happy to be included. Everything runs in your browser through face-api.js on TensorFlow.js. The photo is never uploaded, and recent scans keep only the numbers, never the image.

Want every option? Open the full detector

How to use it

  1. 1

    Pick a sharp group photo where every face is turned roughly towards the camera and is not tiny in the frame.

  2. 2

    Add it to the detector by dropping, browsing, pasting it, or loading an image URL.

  3. 3

    Match each numbered face box on the photo to its numbered chip in the results; Face 1 is the largest face.

  4. 4

    Read the group summary for the face count and the average, youngest and oldest estimate.

  5. 5

    Download the annotated photo with every face boxed and numbered, or copy the plain-text summary to share with the group.

Detecting many faces in one photo

The face detector scans the whole photo and returns a box for every region that looks like a face. A strict pass runs first and keeps only confident detections. If it finds nothing at all, a relaxed pass runs and anything it finds is marked rough. There is no fixed limit on the number of faces; what limits a group photo in practice is face size.

Each face needs enough pixels for the age model to read texture and shape. In a wide group shot, people at the back or the edges can be only a few dozen pixels across, and those are the faces most likely to be missed or to get a shaky estimate. A higher-resolution original, rather than a compressed copy from a chat app, makes a clear difference.

Reading the labelled results

Faces are numbered by size, largest first, and the same numbers appear on the boxes drawn on the photo and on the face chips beside it, so you can always tell which estimate belongs to whom. Tap a chip or a box to see that face's details: the estimated age, the age band, the male or female read with its confidence, and the stability range from the mirrored and cropped re-runs and a photo quality meter for that face.

The group summary pulls the numbers together: how many faces were found, the average estimate, and which face got the youngest and oldest guess. Remember that each figure has its own margin; two people a couple of years apart in the summary may simply be inside each other's range.

Fun ideas for group photos

  • Family line-up: see whether the model can tell parents from grown-up children, and who it thinks looks youngest.
  • Reunion then and now: run an old class photo and a new one, then open Compare to put the two side by side.
  • Friends' guessing game: everyone writes down their guesses before you reveal the AI's.
  • Many separate photos: if you have a folder of portraits rather than one group shot, the Batch tab estimates up to 30 photos in one go.

Getting better results from group shots

  • Use the original photo file rather than a screenshot or a messaging-app copy.
  • Crop a very wide shot so the faces take up more of the frame, then run it again.
  • Even light helps: faces in shadow at the edge of a flash photo are often misread.
  • Faces turned sideways or partly hidden behind someone's shoulder are often missed.
  • Sunglasses, caps and heavy blur all reduce the chance a face is found.

Consent and privacy

A group photo contains other people's faces, so it is worth a moment's thought. Only analyse photos you have the right to use and that the people in them would be comfortable with, and do not share their estimates without asking. The detector itself keeps everything on your device: the photo is analysed in your browser, and the downloaded image is saved only where you choose.

The estimates are guesses from appearance. They are less reliable for children, older adults and some skin tones, and they must never be used to judge anyone's real age, to check whether someone is old enough for something, or to identify anyone.

Frequently asked questions

Can the AI guess the age of everyone in a group photo?

Yes, for every face it can find. Each face gets its own age, age band, gender read and confidence, numbered largest first, with matching boxes on the photo and a summary of the average, youngest and oldest estimate.

How many faces can it detect in one photo?

There is no fixed limit. In practice, faces that are very small, turned away, in deep shadow or partly hidden are the ones that get missed. A higher-resolution photo and a tighter crop help most.

Why was someone in the group missed?

Their face was probably too small, turned sideways, blurred or covered. If the strict pass finds no faces at all, a relaxed pass runs and marks its results rough. Crop closer around the missing person and try again.

Can I analyse lots of separate photos instead of one group shot?

Yes. The Batch tab takes up to 30 photos at once and lists the estimate for the largest face in each, with a CSV download. See the batch age estimation page for details.

Can I save the labelled group photo?

Yes. Download saves the photo with every face boxed and numbered plus a caption marking the figures as estimates, and Copy result puts a plain-text summary on your clipboard.

Is the group photo uploaded?

No. Detection and estimation run in your browser and the photo never leaves your device.

Run a website? Link to this guide if it helps your readers, and see our free backlink sites list for places to get your own site listed.