Face mesh (facial mesh): what it is

A face mesh is the name given to a set of points and lines that describe the surface of the face in an image. A computer vision model looks for the face in the photo and estimates the position of dozens or hundreds of points that follow the eyes, eyebrows, nose, lips and the outline of the face. When those points are joined by lines, a "mesh" forms over the face.

What is the mesh used for in apps?

Once the position of those points is known, the software can do things that would be laborious by hand:

• align two photos by the eyes, to compare before and after;
• locate areas such as the lips, chin or jawline;
• measure distances and simple proportions between points;
• suggest where to place markers for the professional to adjust.

Quality depends on the photo. A face seen from the front, well lit, with no hair in front, no glasses and a neutral expression produces better estimates. Extreme angles, strong shadows and pronounced expressions reduce accuracy. To understand the vocabulary, see also facial landmarks.

What the mesh does not do

It is easy to overestimate the tool. The mesh:

• does not see internal structures, such as muscle, fat, bone or vessels;
• does not diagnose or indicate a procedure;
• does not measure real volume, only the position of points in the image;
• can get it wrong on faces with features the model has seen little of, such as dense beards or marked asymmetries.

Privacy and face data

Facial maps can identify a person and therefore deserve care. In general, this type of data is treated as facial biometric data when it allows the individual to be identified, and the LGPD (Brazil's data protection law) requires a legal basis and transparency. One practice that reduces risk is on-device processing: the analysis happens on the phone and the map does not need to be sent to any server. Under the applicable rules, the patient must be informed about what is collected and why.

In Harmy

The app detects the face on the device itself and shows labeled points over the frontal photo; the professional taps the points for the areas to be treated (Glabella, Temple, Tear Trough, Cheekbone, Nasolabial Fold, Mouth Corner, Lips, Chin, Jawline). The face contours are calculated on the device and are not stored. The photo is sent to a third-party AI only to generate the image, with the professional's consent in the app, and the projection is an illustrative simulation.

Also called: face mesh, facial point mesh.