Face Shapes

How Face Shapes Are Classified

Face shape classification is based on four key measurements, as established in craniofacial anthropometry (Farkas et al., 2005):

1. Forehead width — measured across the widest part of the forehead, typically at the temples.
2. Cheekbone width — the horizontal distance across the most prominent points of the cheekbones.
3. Jawline width — the widest measurement across the jaw, near the angle of the mandible.
4. Face length — the vertical distance from the hairline (or top of the forehead) to the chin.

By comparing the ratios between these measurements, faces can be grouped into categories. Most classification systems in both clinical and beauty contexts recognize six primary types (Milady, 2022). It is common for a person to fall between two categories — face shape exists on a spectrum, not in rigid boxes.

Face Shape Comparison

General Styling Principles by Face Shape

The core principle behind face shape styling is visual balance. Hairstyles, glasses, and makeup are used to either emphasize or counterbalance the dominant proportions of your face (Milady, 2022).

Hairstyles

Hairstylists use face shape as a starting point for recommending cuts. The general approach is to use volume, layers, and framing to create the appearance of an oval silhouette — widely considered the most proportionally balanced shape. For example, round faces benefit from height at the crown and longer layers, while oblong faces benefit from side-swept bangs and volume at the sides.

Glasses and Sunglasses

Eyewear recommendations are based on the principle of contrast: angular frames balance soft, rounded features, while curved frames soften angular structures. The width of the frames should generally match the widest part of the face. Organizations like The Vision Council publish consumer guidelines based on these principles.

Makeup and Contouring

Contouring uses shadow and highlight to sculpt the appearance of facial proportions. The technique varies by face shape: round faces are contoured along the sides to narrow them, square faces are contoured at the jaw angles to soften them, and heart-shaped faces are contoured at the temples to balance the wider forehead.

How AI Detects Face Shape

Traditional face shape identification relies on manual measurement or visual assessment — both of which are subjective and inconsistent. Modern AI approaches use computer vision to detect facial landmarks and calculate proportional ratios automatically.

Our tool uses a Swin Transformer architecture (Liu et al., 2021), a hierarchical vision model that processes images in shifted windows at multiple scales. This allows accurate detection of facial proportions even with variations in lighting, angle, and expression.

The model was trained on labeled datasets of facial images classified by shape. It outputs confidence scores for each of the six categories, so you can see whether you're a clear match or a blend of multiple types. This approach is more consistent than manual assessment and takes seconds rather than minutes.

References

1. Farkas, L.G. (1994). Anthropometry of the Head and Face (2nd ed.). Raven Press.
2. Farkas, L.G., Katic, M.J., & Forrest, C.R. (2005). International anthropometric study of facial morphology in various ethnic groups/races. Journal of Craniofacial Surgery, 16(4), 615-646.
3. Liu, Z., Lin, Y., Cao, Y., Hu, H., Wei, Y., Zhang, Z., Lin, S., & Guo, B. (2021). Swin Transformer: Hierarchical Vision Transformer using Shifted Windows. Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 10012-10022.
4. Milady (2022). Milady Standard Cosmetology (14th ed.). Cengage Learning.
5. The Vision Council. Consumer Guide to Eyewear Selection. thevisioncouncil.org.