Investigation of intelligent barbell squat coaching system based on computer vision and machine learning

2025-03-31

AI Summary

Home-based strength training lacks professional, real-time feedback for squat execution. Method: This study proposes a lightweight, interpretable AI squat coaching system integrating OpenPose-based pose estimation with ensemble classifiers (XGBoost, SVM, RF). It introduces SHAP for interpretable feature selection, focusing on four key biomechanical indicators: joint angles, spinal flexion, knee-to-hip motion ratio, and barbell stability. The system achieves sub-0.5-second per-frame diagnostic latency. Contribution/Results: It achieves F1-scores of 86.86%–100% in detecting six common squat errors, supports real-time corrective feedback and post-session movement replay. User studies demonstrate statistically significant improvements in squat technique (validated by blinded evaluations from both AI and certified strength coaches), confirming the system’s efficacy and practical utility for home-based training.

Technology Category

Humans and AI: Interaction Techniques and DevicesIntelligent Robots: ManipulationMachine Learning: Calibration & Uncertainty Quantification

Application Category

Search and Retrieval-Augmented AI: Web evaluation methodologies and metricsUser Modeling, Personalization and Recommendation: Metrics for user behavior and evaluating successResponsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Purpose: Research has revealed that strength training can reduce the incidence of chronic diseases and physical deterioration at any age. Therefore, having a movement diagnostic system is crucial for training alone. Hence, this study developed an artificial intelligence and computer vision-based barbell squat coaching system with a real-time mode that immediately diagnoses the issue and provides feedback after each squat. In addition, a replay mode allows users to examine their previous squats and check their comments. Initially, four primary characteristics of the barbell squat were identified: body joint angles, dorsiflexion, the ratio of knee-to-hip movement, and barbell stability. Methods: We collect 8,151 squats from 77 participants, categorizing them as good squats and six issues. Then, we trained the diagnosis models with three machine-learning architectures. Furthermore, this research applied the SHapley Additive exPlanations (SHAP) method to enhance the accuracy of issue prediction and reduce the computation time by feature selection. Results: The F1 score of the six issues reached 86.86%, 69.01%, 77.42%, 90.74%, 95.83%, and 100%. Each squat diagnosis took less than 0.5 seconds. Finally, this study examined the efficacy of the proposed system with two groups of participants trained with and without the system. Subsequently, participants trained with the system exhibited substantial improvements in their squat technique, as assessed both by the system itself and by a professional weightlifting coach. Conclusion: This is a comprehensive study that integrates artificial intelligence, computer vision and multivariable processing technologies, aimed at building a real-time, user-friendly barbell squat feedback and training system.

Problem

• Develops AI-based real-time barbell squat coaching system
• Identifies and diagnoses six common squat issues
• Improves squat technique via instant feedback and replay

Innovation

• Computer vision for real-time squat analysis
• Machine learning models for issue diagnosis
• SHAP method for feature selection optimization