02 · PUPILLOMETRY

Pupillometry

A guided 68-second camera experiment for observing relative pupil changes during attention and screen-light tasks.

PUPIL RESPONSE / 4 MINUTES

Enable your camera

A short guided measurement. Camera permission is needed to begin.

01 / 04
Camera setup
Camera preview appears here
Tracking details Check signal before starting
Eye visibility—
Tracking coverage—
Tracking confidence—
Signal size—
Signal stability—
Preparation tips Distance · Light · Comfort
Distance30–40 cm from camera
LightSoft, steady light
ScreenKeep brightness fixed
ComfortBlink normally; stop if uncomfortable

Camera permission is required.

RESEARCH LAB / IMAGE VALIDATION

Measure the measurement.

Upload an eye image, click the pupil center, then mark its left, right, top and bottom edges to draw a reference ellipse. Compare it with the current boundary estimator. Images stay in this browser tab.

For accurate comparisons, upload a close-up of one eye. Processing uses a maximum 480 px image dimension. Results are in resized pixels, not millimeters.

No image selected. Start with a photo from your camera or image library.

Open-source detector comparison

Import image-level CSV or JSON results produced separately by EllSeg, PupilEXT/PuRe, Pupil Labs, or another detector. An independent threshold baseline runs in the browser; named third-party models still require external execution. Columns: image, detector, cx, cy, diameter, width, height. Image coordinates must use the same resized pixel dimensions as the saved reference; mismatches are excluded.

Original projects: PupilEXT · EllSeg · Pupil Labs. Their executables and weights are not bundled. Check their respective licenses and dataset permissions before use.

Exploratory assisted-boundary benchmark. Four-point ellipses are not independent clinical ground truth. No video analysis or automatic image upload is performed. Reloading clears the comparison set.