Golf swing analysis now uses pose estimation models instead of human observation.
Bryson DeChambeau's system applies deep learning on 2D and 3D video frames. It tracks 30 skeletal keypoints across the body, club, and ball. The model labels phases including top of swing, impact, follow-through, and finish.
Practitioners shift from subjective coaching cues to measurable joint angles and club paths. Training plans become data-driven rather than memory-driven. Analysts gain repeatable benchmarks across sessions.
Bryson DeChambeau runs the system on Google Cloud infrastructure. It produces per-swing metrics that guide immediate adjustments during practice rounds.
Step 1: Upload swing footage to Google Cloud Video AI at https://cloud.google.com/video-intelligence. Step 2: Enable pose estimation and export the 30-keypoint JSON file. Step 3: Compare keypoint angles across swings to isolate the largest deviation and adjust grip or stance accordingly.