VIRD: Immersive Match Video Analysis for High-Performance Badminton Coaching.

Area of Science:

• Sports Science
• Human-Computer Interaction
• Computer Vision

Background:

• High-level badminton analysis relies on manual video review, which is time-consuming and lacks effective visualization tools.
• Existing methods create a gap in analyzing matches and communicating insights between coaches and players.
• There is a need for advanced tools to support strategic decision-making in professional badminton.

Purpose of the Study:

• To develop an end-to-end immersive match analysis pipeline for badminton.
• To introduce VIRD (VR Bird immersive analysis tool) for interactive, 3D-based match analysis.
• To enhance the process of gaining insights and developing game strategies for badminton professionals.

Main Methods:

• Designed an immersive analysis pipeline in collaboration with Olympic and national badminton coaches and players.
• Developed VIRD, a VR tool utilizing 3D reconstructed game views for interactive analysis.
• Employed computer vision models to collect and visualize 3D spatial, dynamic shot data, and player poses within the VR environment.

Main Results:

• Immersive analytics in VIRD facilitated effective badminton match analysis.
• Reduced context-switching costs were observed during analysis.
• Enhanced spatial understanding and a high sense of presence were reported by users.

Conclusions:

• VIRD provides an effective solution for interactive badminton match analysis.
• Immersive visualization in VR improves coaches' and players' spatial understanding and strategic insights.
• The tool supports seamless navigation from high-level overviews to detailed rally analysis.