Replication Data for: IMU-based Pelvic Rotation Detection: A Novel Dataset, Benchmark Classifiers, and Sensor Placement Optimization

The dataset contains pseudonymized inertial sensor recordings from 39 healthy adults who performed a structured set of activities in a laboratory environment between February and March 2024 at KU Leuven, Belgium. The activities included sitting, standing, walking, sit-to-stand-to-sit transitions, and targeted pelvic rotations in both anterior–posterior and posterior–anterior directions at varying amplitudes. Data were collected using the entire set of Xsens motion capture sensors at 60 Hz, and activity segments were annotated using synchronized video and the Xsens MVN Avatar software. The aim of the study was to evaluate the feasibility of detecting pelvic rotation and related activities with a single wearable sensor, and to provide a dataset that enables replication of the analyses and further research in activity recognition, posture monitoring, and rehabilitation.