How a face-shape estimate is made
There is no universal mathematical definition of a face shape. MyFaceShaper therefore uses a practical, repeatable set of facial proportions rather than claiming that a photo can reveal one objectively correct label.
What the analyzer sees
When you select a photo, the browser-based facial-landmark model identifies a set of points around the visible face. Those points provide a consistent coordinate system for comparing areas of the face. The current implementation does not need to know your name, age or identity to make the estimate.
The main proportions
The classifier considers overall face width relative to face height, along with relative widths around the forehead, cheekbones and jaw. These values are normalized, so the same basic comparison can be made on a small or large photograph.
| Measurement | Why it matters |
|---|---|
| Overall width / height | Helps separate noticeably elongated outlines from faces that are closer to equal width and height. |
| Cheekbone width | Useful when the cheek area is the visually widest part, as is often described for diamond faces. |
| Jaw width | Helps distinguish a broader, stronger lower face from a jaw that narrows toward the chin. |
| Forehead width | Provides context when comparing the upper face with the cheek and jaw areas. |
Why a rule-based classifier?
The goal of this project is a useful consumer tool, not a research instrument. A rule-based classifier makes the basic reasoning inspectable: the measurements are compared with defined ranges and the result is mapped to one of six familiar categories. It also makes the limitations easier to explain than a black-box claim of certainty.
Why the same person can get different results
A photograph is not a neutral scan of a head. A close camera can change apparent proportions; a tilted head changes the visible jawline; hair can hide the forehead or sides of the face; and lighting can make one side look wider or narrower. Expression can also alter the lower face. These effects are why a borderline result should be treated as a prompt to compare, not as a verdict.
What the categories mean here
Oval, round, square, oblong, heart and diamond are descriptive categories used in common face-shape guides. They are not medical diagnoses and they do not describe every possible facial structure. Real faces can combine characteristics from more than one category.
How to use the result responsibly
If the estimate matches what you see in the mirror, use the corresponding guide for ideas about proportions and styling. If it does not, try a second neutral photograph and compare two likely categories. The useful outcome is better understanding, not forcing your face into a label.
Try the method
Upload a clear front-facing photo and use the estimate as a starting point for exploring the guides.
Open the analyzer