IMU Head Movement Analysis
About Project
- Type: Wearable Sports Science Analytics (IMU-Based)
- Data: IMU time-series (gyroscope, accelerometer, angles, and quaternions when available)
- Deliverables: Per-athlete summary outputs + cohort-level CSV summaries + visual plots for coach review
- Use Cases: Movement quality screening, stability monitoring, consistency tracking, and training focus recommendations
Objective
Convert raw IMU recordings into coach-readable insights by detecting movement events, quantifying head-movement intensity and stability, and producing consistent summaries that highlight movement quality and potential training focus areas.
Tools & Technologies
Python, Pandas, Time-Series Cleaning, Basic Signal Processing (Smoothing / Filtering), Event Detection (Windowing / Peak-Based Segmentation), Feature Engineering, Visualization, CSV + Report Outputs
Key Work & Impact
Cleaned and aligned multi-sensor IMU streams to handle real-world issues (missing values, inconsistent recordings, variable session lengths, and noisy segments).
Built a repeatable movement-event detection approach using smoothed gyro magnitude and window-based segmentation to isolate individual movement bouts from continuous recordings.
Engineered athlete-friendly movement metrics such as peak rotational intensity, sustained-intensity exposure, range-of-motion proxies, and stability/consistency indicators across detected events.
Generated visual diagnostic plots (trend lines + detected movement windows) so coaches can validate patterns quickly instead of trusting a “black box” score.
Produced per-athlete summaries and a cohort-level export (CSV) to enable comparisons across athletes, sessions, and time periods.
Designed the workflow for iteration: metrics and thresholds can be tuned based on sport context, drill type, and coach feedback without rebuilding the pipeline.