Real-time Fencing Action Recognition with Mediapipe

Real-time Fencing Action Recognition with Mediapipe

This paper investigates the use of Mediapipe for real-time fencing action recognition. Mediapipe, a powerful open-source framework for building machine learning pipelines, offers efficient and accurate solutions for various tasks, including pose estimation and gesture recognition.

The proposed system leverages Mediapipe's capabilities to detect key points on the fencer's body, creating a representation of their pose. This pose information is then fed into a machine learning model trained to distinguish different fencing actions, such as lunges, parries, and attacks.

The paper explores the following aspects:

• Accuracy of Mediapipe in detecting key points for fencing actions.
• Performance of different machine learning models for action classification.
• Real-time processing capability of the system.
• Potential applications in sports analytics and training.

The research aims to demonstrate the effectiveness of Mediapipe for real-time fencing action recognition, contributing to the development of innovative tools for sports analysis and training. The system's capabilities can be extended to other sports with similar action patterns, showcasing Mediapipe's versatility and potential for wider applications in sports technology.