Cerv-D: A Dataset of Cervical Deviations During Shoulder Exercises Derived from 3D Kinematic Analysis

Cerv-D measures cervical (neck) deviations during shoulder exercises using 3D kinematic data obtained from pose estimation and landmark tracking (Shoulder-KX Dataset). The process begins by identifying key anatomical landmarks, including the nose and the left and right claws (located in the foot). The midpoint between the two claws is calculated and a line is drawn from the nose to this midpoint. The angle between this line and the vertical axis is then determined, representing the tilt of the head and neck. The neck deviation is calculated as the difference between this angle and the ideal upright posture of 90 degrees; this deviation quantifies how much the head is misaligned during the exercise.