Framework for Intelligent Swimming Analytics with Wearable Sensors for Stroke Classification.

Area of Science:

• Sports Science
• Data Analytics
• Wearable Technology

Background:

• Optimizing athlete training and performance through data analysis is crucial in high-performance sports.
• Intelligent algorithms and wearable hardware advancements enable real-time pattern recognition for sports coaching.

Purpose of the Study:

• To propose an intelligent data analytics system for swimmer performance evaluation.
• To leverage wearable sensors and advanced algorithms for real-time swimming analytics and coaching support.

Main Methods:

• Development of a system for pre-processing raw signals from wearable inertial (AHRS) and biosensors (heart rate, pulse oximetry).
• Implementation of online recognition for swimming style and turns, alongside post-analysis for stroke counting and average speed.
• Utilizing radio-frequency links for data communication and analytics in a real training environment.

Main Results:

• The system effectively provides immediate feedback to coaches and swimmers through real-time data analysis.
• The Random Forest classifier achieved a high accuracy of 95.02% (macro-averaged F1 score).
• The framework demonstrated significant benefits in supporting coaches monitoring multiple swimmers during training.

Conclusions:

• The proposed intelligent swimming analytics system with wearable sensors is effective for real-time performance feedback.
• The system enhances coaching decision support by providing detailed insights into swimmer performance.
• This technology represents a significant advancement in high-performance sports coaching and athlete development.