Computer vision and CAD · 2022–2023
KeyCAD
An image-to-CAD pipeline for parameterized key models.
Overview
A computer-vision and CAD experiment that translated measurements from photographs into dimensionally accurate, 3D-printable key geometry.
Challenge
Turning a flat photograph into useful physical geometry required a consistent way to correct scale, locate measurement points, and map those measurements onto a precise 3D model.
Approach
I separated the system into two parts: a library of parameterized OpenSCAD blanks and a Python/OpenCV measurement pipeline that selected and modified the matching model.
Result
The finished pipeline connected image analysis and procedural CAD into a repeatable end-to-end workflow across 19 key profiles.
Highlights
- Created 19 parameterized key-blank models from manufacturer specifications.
- Built an OpenCV pipeline that measured positions and depths from photographs.
- Generated dimensionally accurate STL geometry from the corresponding parameterized model.