research
A High Agreement Score Hid Millimeters of Error
A low-cost facial depth sensor ranked people consistently but missed the precision needed for reconstruction.
Summary
A low-cost facial depth sensor ranked people consistently but missed the precision needed for reconstruction.
Two observers measured 12 adults at 16 facial distances using a low-cost depth sensor, a structured-light scanner and direct calipers. Device-versus-caliper agreement looked high by intraclass correlation, at 0.88 and 0.93, yet mean absolute error was 2.75 millimeters versus 1.64 millimeters for the comparator and 14 of 16 depth-sensor distances exceeded a two-millimeter tolerance. The authors conclude that the inexpensive device suits surveys, ranking and within-person monitoring, not high-fidelity reconstruction; high correlation alone did not establish clinical usability.
Why it matters
A low-cost facial depth sensor ranked people consistently but missed the precision needed for reconstruction.
Limits and context
- The authors conclude that the inexpensive device suits surveys, ranking and within-person monitoring, not high-fidelity reconstruction; high correlation alone did not establish clinical usability.
Key claims
A low-cost facial depth sensor ranked people consistently but missed the precision needed for reconstruction.
Qualification: The authors conclude that the inexpensive device suits surveys, ranking and within-person monitoring, not high-fidelity reconstruction; high correlation alone did not establish clinical usability.
Evidence: source-2026-08-04-006
Sources
- Journal of Craniofacial Surgery: low-cost facial depth-sensor validationWolters Kluwer Health · secondary reporting
Corrections
No corrections have been recorded for this story.