Comparative Analysis of Marker-based vs. Marker-less Computer Vision for a Gamified Lower Limb Rehabilitation

Abstract

Gamified computer vision systems offer low-cost alternatives to supervised rehabilitation, yet direct comparisons of marker-based and marker-less approaches using identical hardware remain limited. This study compared two ankle rehabilitation systems using the same webcam (eMeet C960): a custom HSV-based marker tracking system and a marker-less system based on MediaPipe Pose. Both were integrated into a Unity game targeting dorsiflexion, plantarflexion, inversion, and eversion. Ten healthy participants evaluated the systems using a 5-point Likert questionnaire. The marker-based system achieved lower angular error (3.2° vs. 7.16° ± 1.21%) and higher usability ratings, particularly for standing play and elderly users. These findings suggest marker-based tracking is better suited for clinically precise rehabilitation, whereas marker-less tracking offers greater accessibility and ease of use for home-based applications.