The purpose behind the pixels
A helping hand.
A human purpose.
A student-led passion project exploring a simpler way to size 3D-printed prosthetic hands.
More accessible, from the start.
Getting a prosthetic hand involves people, time and careful fitting. For growing children, that process may need to happen again as their needs change.
AI HandSizer brings machine learning and 3D printing together to make the initial measurement step easier, especially for families and volunteer makers with limited resources.
Our aim is practical: help people move from a photo to a useful conversation about fit.
Where we are today
A research tool, built to be useful.
Current sizing uses the bundled e-NABLE Phoenix table and a photograph of an intact hand. Model suitability and final fitting need individual review by your maker or prosthetist.
Additional hand models are a future ambition, not a feature available today.
Behind your result
From a photo to a print scale.
- 01 · DetectFind the hand and standard-size reference card.
- 02 · CalibrateUse known standard-size card dimensions to convert pixels to millimetres.
- 03 · MeasureEstimate palm width and hand length.
- 04 · MatchMap dimensions to a Phoenix print scale.
Workflow: a standard-size reference card provides the pixel-to-millimetre scale. Quality checks flag photos that need to be retaken.
Honest about accuracy
An estimate.
Never a promise.
The current calibration uses a dataset-level median combined measurement error of approximately 11.72 mm. This is not an accuracy guarantee, a confidence percentage, or an error bound for your particular photo.
Broader real-world validation and a fuller public evaluation are still needed.
Help shape what comes next.
Was the guide easy to follow? Did the measurement help a maker? Your experience can tell us what needs improving.
Share your experience →