Smartphone Positioning Affects Pose Estimation Accuracy

This cross-sectional within-subject study of 44 university students evaluated how smartphone camera angle (front, side, diagonal) and distance (90, 180, 200, 360 cm) affect AI-based pose-estimation detection and repetition counting for push-ups and squats. Across ~1,320 repetitions per exercise, mean detection rates were ≈61% with MAEs ≈1.1 repetitions; diagonal and frontal mid-range views (180–200 cm) produced the highest accuracy. Findings offer actionable guidance for mHealth developers and clinicians.