Deep learning now labels every frame of your golf swing from phone video
Bryson DeChambeau's system runs proprietary 2D and 3D deep-learning models on standard video. It tracks more than thirty keypoints across body, club, and ball through top of swing, impact, follow-through, and finish. Shot outcomes are logged automatically for each sequence.
Players replace subjective coaching cues with frame-by-frame kinematic data. Training decisions move from opinion to statistical comparison across sessions. The same phone becomes both camera and performance lab.
Bryson DeChambeau publishes swing metrics on the Google Cloud blog and applies the models during practice rounds. His team reports consistent reductions in dispersion after using the labeled keypoint data.
Step 1: Open Google Cloud's AI video tools or DeChambeau's published pipeline at cloud.google.com. Step 2: Upload a swing video and run the 2D/3D pose model to extract the thirty-plus keypoints. Step 3: Export the labeled timeline, identify the largest angle deviation at impact, and adjust grip or stance before the next recording.