Happy new year!
After re-framing the putting as a time-series problem, the next question was how you can capture the motion.
Most putting studies rely on motion capture systems or sensor platforms that are precise but impose constraints such as fixed environments, limited accessibility, and data that’s hard to reproduce outside the lab. This didn’t align with what I wanted, as the goal was to understand putting motion in a flexible, repeatable way.
I decided to use a camera-based approach using high-frame-rate videos from phones and visual (ArUco) markers. I placed the marker on the putter and recorded from camera views synchronized using a flash, and I reconstructed the putter and person’s motion frame by frame, capturing position, rotation, etc.
This preserved the temporal continuity of the stroke, and it make the system scalable. The setup could be reproduced with accessible equipment, so the analysis wouldn’t depend on a single controlled environment.
By this stage, I wasn’t trying to prove “good” or “bad” putting but to make a consistent framework that reflect how small movements like putting.