Loading page
About this tool
Upload a photo and our advanced AI will detect the person's estimated age and gender in seconds.
The AI Age & Gender Detector reads a photo and returns an estimated age, the age band that estimate falls in, and a male/female classification with the confidence behind it — one set of numbers per face. Detection and inference both run inside this page through face-api.js on TensorFlow.js, so the photo is analysed on your own phone or laptop and never reaches a server. It is a guess from visible features, useful for curiosity and for seeing how a vision model reads a portrait — never for age verification or identity.
Open Age & Gender Detector on AltFTool — it loads instantly in your browser.
Drop a JPG, PNG or WebP photo onto the upload panel, or tap Choose Photo to pick one from your phone gallery.
Prefer a live shot? Tap Use Camera, switch between the front and rear lens, then hit Capture Photo.
Wait once for the face-api.js model (about 620 KB) to reach your browser cache, then read the estimated age and the age bracket it falls in.
Check the gender label with its confidence bar, and the detection score that says how sure the detector was it found a face at all.
In a group shot, tap the Face 1 / Face 2 chips to move between everyone the detector found, largest face first.
Press Copy result for a plain-text summary, or Download image to save the photo with every face boxed and labelled.
Face detection and the age/gender estimate both run in your browser through TensorFlow.js. Nothing is uploaded, nothing is stored, and there is no server copy to delete afterwards.
Group photos return a separate age and gender for every face found. Numbered boxes on the photo line up with the Face chips, so you always know which reading belongs to whom.
Gender confidence and a separate face detection score are both shown, so a strong read is visibly different from a weak one instead of both arriving as a bare number.
If the strict detector finds nothing, a relaxed pass runs automatically and the result is flagged as rough — far better than a dead-end 'no face detected' on an off-angle or dimly lit photo.
The tiny face detector and the age/gender network are fetched on your first analysis and then served from your browser cache, so later photos start immediately.
Copy result puts a plain-text summary on your clipboard; Download image saves the photo with every face boxed and labelled, plus a caption strip carrying the headline numbers and a one-line reminder that this is an estimate. No account and no daily limit.
Treat it as a guess with a wide margin, not a measurement. It lands closest on a sharp, evenly lit, front-facing photo of an adult, and drifts furthest on children, heavy makeup, beauty filters, strong side lighting and off-angle shots. That is why the tool shows a face detection score next to the estimate — a weak score means the age came off a poor crop and should be discounted. No online tool can honestly promise an accuracy figure for your photo, and this one does not.
No. The face detector and the age/gender network are downloaded into your browser and run there. Your photo is read from the file or camera straight into the page, analysed on your device, and released when you press Analyze another or close the tab. There is no upload step, so there is no copy on AltFTool's servers to store or delete.
No. It estimates age from appearance and can be off by a decade; it can also be fed a photo of a photo. It cannot tie a face to a person or a document. Never use it to grant or refuse access, to decide whether someone is old enough for something, or as evidence about anyone.
No, and it was never built to. The model has exactly two outputs and classifies visible facial features against how its training data was labelled. Gender identity is not visible in a photograph. Please don't use the output to make assumptions about a real person.
The detector runs twice — once strict, then once relaxed — and only reports a miss when both come up empty. In practice that means the face is small in the frame, turned too far to the side, backlit by a window, or partly covered by sunglasses or a cap. Move closer, turn towards the light, uncover the forehead and eyes, and try again.
Yes. Every face the detector finds gets its own age, band, gender and confidence. Results are ordered largest face first, so Face 1 is normally the subject rather than a passer-by in the background, and the numbered boxes drawn on the photo match the Face chips above it.
The first analysis downloads about 620KB of model files — a tiny face detector plus the age and gender network. Your browser caches them, so every photo after that skips the download and runs only the analysis, which is where the big speed difference comes from.
Any image file up to 15MB is accepted, and the decoding is then done by your browser itself — so JPG, PNG, WebP and GIF work everywhere. HEIC is the exception: it opens only where the browser can decode HEIC, which in practice means Safari on Mac and iPhone. Hand the same .heic file to Chrome, Edge or Firefox and the tool reports that it could not be opened as an image, so export or re-save it as JPG first. Otherwise it runs the same way in any modern mobile or desktop browser: pick a photo, drag one in, paste one with Ctrl / ⌘ + V on a desktop, or use the phone's front or rear camera.
Yes — free, with no signup and no daily limit on how many photos you analyse. The image you download carries the face labels and a short caption line noting the numbers are an estimate; there is no other branding, paywall or export restriction.