Computer Vision for Movement Analysis: Pose Estimation Technologies for Assessing Exercise Form, Gait, and Rehabilitation Progress Without Laboratory Equipment

Abstract

Introduction and Objective: Laboratory-based motion capture systems provide highly accurate 
measurements of human movement but require specialized equipment, controlled environments, and trained 
personnel. Recent advances in computer vision and markerless human pose estimation (HPE) have created 
opportunities to assess movement using ordinary cameras. This review examines the application of 
pose-estimation technologies, particularly OpenPose and MediaPipe, for exercise-form assessment, gait 
analysis, and rehabilitation progress monitoring, with emphasis on measurement accuracy, reliability, and 
practical limitations. 
Materials and Methods: A narrative review of peer-reviewed literature was conducted focusing on 
markerless human pose estimation, computer-vision-based movement analysis, OpenPose, MediaPipe, gait 
analysis, exercise monitoring, and rehabilitation. Studies comparing pose-estimation-derived measurements 
with conventional marker-based motion-capture systems were considered particularly relevant. Evidence 
was synthesized according to technical characteristics, movement-analysis applications, measurement 
agreement, and factors affecting real-world implementation. 
Results: The reviewed literature indicates that markerless pose estimation can provide useful estimates of 
joint positions, movement trajectories, and spatiotemporal gait parameters without laboratory-grade 
instrumentation. Validation studies have reported good-to-excellent agreement between markerless systems 
and conventional motion capture for many temporal gait variables, while accuracy tends to be lower for 
fine-grained joint-angle measurements and small-amplitude movements at distal joints such as the ankle. 
Performance is also influenced by camera position, lighting, clothing, occlusion, movement direction, and 
the dimensionality of the camera system (Cao et al., 2019; Hii et al., 2023; Stenum et al., 2021). 
Conclusion: Computer-vision-based pose estimation represents a promising and accessible approach for 
movement screening, exercise-form monitoring, gait assessment, and rehabilitation progress tracking. 
However, current systems should generally be regarded as complementary tools rather than complete 
replacements for laboratory-grade motion-capture systems when high-precision biomechanical measurement 
is required.