Face Comparison Tool

The tool runs face-api.js entirely in your browser. It detects the face in each photo, extracts a 128-dimensional face descriptor — a compact numerical representation of facial geometry — then computes the Euclidean distance between the two descriptors. The distance is converted into a similarity percentage: a lower distance means a higher, closer match.

No. All processing runs locally in your browser using face-api.js. Neither photo is ever sent to a server. The only external request is to download the AI model files (once, cached afterward). Once you close the tab, nothing is retained.

Scores of 80%+ indicate extremely close structural similarity — often seen with identical twins or the same person in different photos. 60–79% suggests a strong resemblance. 40–59% is moderate similarity with clear individual differences. Below 40% means the two faces have distinctly different facial geometry.

Yes — this is one of the most useful ways to use the tool. Comparing two photos of yourself (e.g., a childhood photo vs. a recent one) shows how much your facial structure has changed over time, since bone structure ratios tend to stay more stable than skin, hair, or weight.

The AI needs a clear, front-facing photo with the full face visible and reasonably even lighting. Extreme angles, heavy shadows, sunglasses, or multiple people in frame can prevent detection. Try a different, clearer photo.

This tool uses face-api.js, which runs three neural networks in-browser: a face detector, a 68-point landmark model, and a face recognition model that outputs a 128-dimensional descriptor. Comparing descriptors via Euclidean distance is the same underlying approach used in professional face recognition systems.

Yes, to a degree. The model is trained to be robust to lighting and expression changes, but extreme differences — heavy shadows, exaggerated expressions, different angles — can shift the score. For the most accurate comparison, use two clear, neutral-expression, front-facing photos.