Action Recognition of Taekwondo Unit Actions Using Action Images Constructed with Time-Warped Motion Profiles

Area of Science:

• Sports Science
• Biomechanical Engineering
• Computer Vision

Background:

• Taekwondo is an Olympic sport demanding precise and rapid movements.
• Recognizing Taekwondo actions is challenging due to complex kinematics.
• Wearable inertial measurement unit (IMU) sensors offer a viable solution for capturing motion data.

Purpose of the Study:

• To develop a novel action recognition model for Taekwondo unit actions.
• To leverage joint-motion data from IMU sensors for enhanced motion capture.
• To improve the accuracy and effectiveness of Taekwondo technique recognition.

Main Methods:

• Utilized a convolutional neural network (CNN)-based image classification framework.
• Synthesized action images by mapping joint-motion profiles onto the RGB color space.
• Applied a time-warping technique to refine rapid movement representation based on action velocity.

Main Results:

• Achieved high performance metrics on a dataset of 40 Taekwondo experts.
• Reported an accuracy of 0.998, precision of 0.983, recall of 0.982, and F1 score of 0.982.
• Demonstrated the effectiveness of the time-warping technique in enhancing feature representation.

Conclusions:

• The proposed model effectively recognizes Taekwondo unit actions.
• The method shows scalability and robustness in real-world applications.
• This approach advances the field of sports biomechanics and action recognition technology.