MultiSenseBadminton
Scientific Data · 2024 · Lead author
How can we study a stroke beyond its outcome?
A successful shot alone does not describe the movement, muscle activity, or gaze behind it. MultiSenseBadminton brings these signals together for analysis across skill levels.
Dataset and approach
The dataset contains 7,763 strokes from 25 players, covering forehand clears and backhand drives. It combines body motion, gaze, muscle activity, foot pressure, video, and annotations. Three professional coaches informed its design.
Evidence and limits
The published study reports technical validation and preliminary classification benchmarks. These demonstrate analysis possibilities, not improvements in learning. Two strokes in a launcher-based setting, sensor noise and missing recordings, subjective labels, and class imbalance limit generalization to match play.
My contribution
As stated in the paper’s author contributions, I conceived and conducted data collection, preprocessed the dataset, performed technical validation, and drafted the manuscript.