Face Shape Analyzer

How this face shape analyzer works

This tool runs Google's MediaPipe FaceLandmarker model directly in your browser to find 478 facial landmark points in your photo — the edges of your forehead, cheekbones, jawline, chin and more — then computes simple geometric ratios between them: face length versus the widest part of your face, jaw width versus cheekbone width, forehead width versus cheekbone width, and how much the jaw narrows toward the chin. Those ratios are compared against simple thresholds to estimate one of seven shape categories. The full landmark list and thresholds are documented on the about page — nothing is a black box.

The seven face shape categories

• Oval — Balanced proportions: length is somewhat greater than width, with forehead, cheekbones and jaw of similar width. Most frame shapes and hairstyles suit oval faces well — it's generally considered the most flexible shape to style for.
• Round — Face length and width are close to equal, with a soft, rounded jawline. Angular or rectangular frames, and hairstyles with height or sharper lines, are often used to add definition and visually elongate a round face.
• Square — Face length and width are close to equal, with a strong, angular jawline about as wide as the cheekbones and forehead. Round or oval frames, and softer, layered hairstyles, are often used to balance a strong jawline.
• Heart — Forehead is the widest area, narrowing toward a smaller, sometimes pointed chin. Frames wider at the bottom, and hairstyles with volume near the jaw, are often used to balance a wider forehead.
• Triangle — Jawline is the widest area, narrowing toward a smaller forehead. Frames wider at the top, and hairstyles with volume near the temples, are often used to balance a wider jaw.
• Diamond — Cheekbones are the widest area, with a narrower forehead and jaw. Oval or rimless frames, and hairstyles that add width at the forehead or jaw, are often used to soften prominent cheekbones.
• Oblong — Face length is notably greater than width, with fairly similar forehead, cheekbone and jaw widths. Frames with more lens depth and hairstyles with width at the sides are often used to visually shorten an oblong face.

Why nothing gets uploaded

The face-landmark model — a compact neural network published by Google as part of MediaPipe — is downloaded once to your device, and every photo you analyze after that is processed locally by your browser's own GPU (via WebGL/WebGPU) or CPU (via WebAssembly). There is no server-side image processing on this site at all, which means there is no server log, no storage bucket, and no possibility of your photo being seen by anyone else. The photo lives only in a JavaScript variable and an on-page canvas for as long as the tab is open — it's discarded the moment you upload a new photo, close the tab, or refresh the page.

FAQ

Is this a scientific or medical measurement?

No. It's a heuristic geometric classification based on a handful of pixel-distance ratios from one photo. It has no clinical or cosmetic-industry validation behind its category boundaries — they're descriptive labels chosen for readability, not established standards.

Does my photo get uploaded anywhere?

No. Facial landmark detection runs entirely in your browser using MediaPipe's WebAssembly/WebGL runtime. See the privacy policy for the full explanation.

Why did it say "no face detected"?

The model needs a reasonably clear, front-facing, well-lit view of a face. Extreme angles, heavy occlusion (masks, hands, hair fully covering the face), very low light, or a face that's too small in the frame can all cause detection to fail.

Why might the result seem wrong?

The measurements are sensitive to head tilt, camera angle, hairstyle (hair covering your forehead or jawline shifts where the model finds those edges), and facial expression. Face shape is also inherently a simplification — real faces blend categories rather than falling cleanly into one.

Are the style suggestions personalized advice?

No. They're generic, commonly-cited style heuristics for each shape category, not personalized recommendations from a stylist or optician, and this site has no affiliate or product relationships tied to them.

Informational and entertainment tool only — not a substitute for professional styling or optical advice.