Detecting and Visualizing Stops in Dance Training by Neural Network Based on Velocity and Acceleration

Area of Science:

• Dance studies
• Biomechanical analysis
• Human-computer interaction

Background:

• Traditional group dance practice, including Yosakoi Soran, promotes community and health.
• COVID-19 restrictions have significantly limited face-to-face group dance activities.
• There is a growing need for effective remote dance practice solutions.

Purpose of the Study:

• To propose and evaluate a system for detecting and visualizing critical dance movements, specifically 'stops'.
• To leverage technology to facilitate remote group dance practice and maintain community engagement.
• To explore the application of motion capture and AI in dance analysis.

Main Methods:

• Utilizing motion capture technology to record dance movements.
• Calculating movement features based on velocity and acceleration data.
• Employing a neural network to learn motion features for stop detection.
• Visualizing detected stops using a human-like 3D model.

Main Results:

• The proposed system achieved high accuracy in detecting dance stops.
• The visualization component effectively represents critical dance moments.
• Experimental results validated the system's effectiveness for remote practice.

Conclusions:

• The developed system offers a viable solution for remote group dance practice.
• Information and communication technology can effectively support dance communities.
• Accurate detection and visualization of dance movements enhance remote training.