How accurate is the smile analysis?

Smile Tracker uses Google MediaPipe Face Landmarker, which maps 478 facial landmarks and reads 52 blendshape scores. Accuracy is highest with clear, well-lit, front-facing photos. Results are estimates based on facial geometry, not clinical measurements.

Is my photo stored or uploaded anywhere?

No. All analysis runs entirely inside your browser using client-side JavaScript. Your photo is never uploaded to any server, database, or third-party service.

What is a Duchenne smile?

A Duchenne smile is a genuine smile that activates both the zygomatic major muscle (mouth corners) and the orbicularis oculi (around the eyes). Named after neurologist Guillaume Duchenne, it is the most authentic and attractive kind of smile.

What makes a smile score high vs low?

The AI measures mouth smile curve, cheek lift, eye squint (Duchenne marker), and jaw openness. A high score means strong, genuine happiness across all four signals. A low score may simply mean the photo didn't capture your best moment.

What photo gives the best results?

A clear, front-facing photo with good natural lighting works best. Avoid heavy shadows, extreme angles, or sunglasses. A genuine, relaxed expression produces more meaningful results than a stiff pose.

← Back to Smile Tracker

About Smile Tracker

A free, private AI tool for understanding the science behind your smile.

What Is Smile Tracker?

Smile Tracker is a free, browser-based smile analysis tool that uses Google MediaPipe Face Landmarker to analyze the authenticity, intensity, and type of your smile from a single photo. No account. No upload. No server.

Every analysis runs entirely inside your browser using client-side JavaScript. Your photo is processed locally on your device and is never transmitted anywhere.

The Technology

Smile Tracker is powered by Google MediaPipe Face Landmarker, an open-source machine learning framework used across Google products including Google Meet, YouTube, and Pixel cameras.

The model maps 478 unique 3D facial landmark coordinates and produces 52 blendshape coefficients representing individual facial muscle activations. These values quantify signals like mouth smile curve, cheek lift, eye squint, and jaw openness.

These four core signals are fed into a weighted scoring formula to calculate your Smile Score (0–100) and classify your smile type.

The Duchenne Difference

The central insight behind Smile Tracker comes from 19th-century neuroscience. In 1862, French neurologist Guillaume Duchenne identified that genuine smiles require two muscle groups simultaneously — the zygomatic major (mouth) and the orbicularis oculi (eyes).

Because the eye muscle is controlled by a different neural pathway than voluntary muscles, the "eye smile" cannot be consciously faked. Smile Tracker detects this Duchenne marker using MediaPipe's eye squint blendshapes and uses it as a primary signal for genuine smile classification.

Privacy Commitment

We built Smile Tracker with a privacy-first architecture by design, not as an afterthought.

No photo is ever uploaded to any server.
No personally identifiable information is collected.
No account creation is required.
Analysis runs fully offline after the initial page load.
Your photo is discarded immediately when you leave or reset the tool.

Contact

Questions, feedback, or press inquiries? Reach us at thesmiletracker@gmail.com.

Analyze My Smile Free →