Face Shape Data
Everything the face shape detector uses to turn a photo into one of seven shapes, published as plain files so anyone can check it, copy it or argue with it.
What is in here
- face-shape-ratios.csv — the seven target profiles: for each shape, the three ratios the detector compares against, and how far a ratio may sit from the target before it stops counting as a match.
- face-shape-measurements.csv — the four measurements, the landmark indices each one is taken between, and how each is reported.
- face-shape-worksheet.pdf — the printable tape-measure worksheet, as a PDF.
- golden-ratio-ratios.csv — the five ratios the golden ratio face check reads, with the target and tolerance for each one and where the target comes from.
- golden-ratio-landmarks.csv — every landmark index used on that page and what each one marks.
- golden-ratio-formula-sheet.pdf — the same ratio table and formulas as a printable sheet.
How the numbers are produced
- A face landmark model finds 468 points on the photo. The model file is served from this site and runs on your own device; the photo is never uploaded.
- Four widths and lengths are read off those points: forehead width, cheekbone width, jaw width and face length.
- All four are expressed as a percentage of face length, which is fixed at 100. They are relative proportions of your own photo, not millimetres or inches.
- Three ratios follow: length ÷ cheekbone width, forehead width ÷ cheekbone width, and jaw width ÷ cheekbone width.
- Each of the seven target profiles is scored by how far those three ratios sit from it, scaled by the tolerances in the CSV. The closest profile wins; the score shown in the tool is that closeness measure, not a probability.
The seven target profiles
| Shape | length ÷ cheekbone | forehead ÷ cheekbone | jaw ÷ cheekbone |
|---|---|---|---|
| Oval | 1.45 | 0.86 | 0.76 |
| Round | 1.18 | 0.90 | 0.86 |
| Square | 1.22 | 0.97 | 0.95 |
| Heart | 1.42 | 1.00 | 0.70 |
| Diamond | 1.50 | 0.74 | 0.72 |
| Oblong | 1.75 | 0.88 | 0.84 |
| Triangle | 1.32 | 0.80 | 1.02 |
A ratio may sit up to 0.30 (length ÷ cheekbone) or 0.12 (the two width ratios) away from a profile and still count as a match, which is why two shapes can score close together on the same face.
What this data cannot tell you
- A photo has no scale reference, so no tool can give you real-world millimetres or inches from a photo alone. These are proportions of your own image.
- The profiles are a styling estimate, not a medical or anatomical classification, and not a measurement of a real frame size.
- Pose, lighting, hair across the forehead, glasses and lens distortion all shift the landmark points, and therefore the ratios.
- Someone can sit between two profiles. The tool reports the closest one; it does not claim the others are wrong.
Using or citing this data
If you want to refer to these numbers, this wording is accurate: “Target ratio profiles and tolerances published by Face Shape Hub (faceshapehub.com/data/), retrieved [date].”
The same files are also archived with a DOI, so the citation line above can point at a frozen version: 10.5281/zenodo.22823412 (Zenodo). The record lists the files, their sizes and the version history.
Last revised 20 September 2026. The date above it is the archived version the DOI points to.
These files are published under the Creative Commons Attribution 4.0 International licence (CC BY 4.0): you may copy, redistribute and adapt them, including commercially, as long as you give appropriate credit — the citation line above is enough. The licence text is the authoritative version.