Dance of Fireworks: An Interactive Broadcast Gymnastics Training System Based on Pose Estimation

TL;DR

This work introduces Dance of Fireworks, a mobile, camera-based interactive training system that uses lightweight pose estimation (PoseNet/OpenPose) to guide office workers through radio calisthenics. By computing joint angles and comparing them to demonstrations via a cosine-similarity framework with a 3-second delay, the system provides real-time feedback and maps motion quality to fireworks-based rewards, enabling engaging, hardware-light deployment. In a study with 136 participants, the approach achieved a significant reduction in average joint-angle error from $21.3^ ext{°}$ to $9.8^ ext{°}$ ($p<0.01$) over four sessions and received strong user endorsement for both exercise promotion and entertainment value. The work demonstrates a cost-effective, scalable solution for reducing sedentary risk in office environments and outlines future enhancements including improved pose accuracy, lower latency, multiplayer features, music synchronization, and 3D pose capabilities.

Abstract

This study introduces Dance of Fireworks, an interactive system designed to combat sedentary health risks by enhancing engagement in radio calisthenics. Leveraging mobile device cameras and lightweight pose estimation (PoseNet/TensorFlow Lite), the system extracts body keypoints, computes joint angles, and compares them with standardized motions to deliver real-time corrective feedback. To incentivize participation, it dynamically maps users' movements (such as joint angles and velocity) to customizable fireworks animations, rewarding improved accuracy with richer visual effects. Experiments involving 136 participants demonstrated a significant reduction in average joint angle errors from 21.3 degrees to 9.8 degrees (p < 0.01) over four sessions, with 93.4 percent of users affirming its exercise-promoting efficacy and 85.4 percent praising its entertainment value. The system operates without predefined motion templates or specialised hardware, enabling seamless integration into office environments. Future enhancements will focus on improving pose recognition accuracy, reducing latency, and adding features such as multiplayer interaction and music synchronisation. This work presents a cost-effective, engaging solution to promote physical activity in sedentary populations.

Figures (13)

• Figure 1: The screen cap of Just-Dance
• Figure 2: The screen cap of Active5
• Figure 3: The pipeline of the system.
• Figure 4: Keypoints in COCO keypoints challenge.
• Figure 5: The pipeline of Openpose. Image taken from 8765346