This study investigates the complex interplay between movement patterns and cognitive/emotional states to gain a deeper understanding of athlete performance and psychological well-being. To overcome the limitations of traditional analytical methods, this research employs state-of-the-art Computer Vision (CV) techniques to accurately capture, track, and analyze athletes’ fine motor and gross movements in training and competition environments.
The paper first details how deep learning-based skeleton estimation algorithms can be utilized to perceive human motion. Dynamic movement data—such as an athlete’s posture, acceleration, deceleration, and trajectory changes—is extracted to build a comprehensive Movement Pattern Dataset in 3D space. This movement data is then correlated with psychological state indicators, including the athlete’s stress level, motivation, self-confidence, and decision-making speed. This correlation is established using standardized psychometric assessments and biometric sensor data (e.g., heart rate variability) provided by coaches and sports psychologists.
The core novelty of the work lies in the development of a hybrid Deep Neural Network (DNN) architecture capable of predicting psychological states from movement patterns. This model takes movement data as input and estimates the athlete’s current psychological profile. The results indicate that specific movement variations or anomalies show a strong correlation with critical psychological shifts, such as high stress or low self-confidence.
Potential applications of this modeling approach include real-time performance feedback, proactive prediction of injury risk based on psychological factors, and the design of personalized training regimens. In conclusion, this research demonstrates the potential of Computer Vision technology to revolutionize sports science and psychology, paving a new way for objective and data-driven decision-making.