Researchers have developed an event-based active vision system designed to accurately estimate the spin of balls in real-time across various professional sports. This system utilizes an event camera for high temporal resolution, galvanometer mirrors for continuous tracking, and a focus-tunable telephoto lens for enhanced detail. An offline method, s-CMax, achieves state-of-the-art accuracy on static balls, while a low-latency online method, employing a convolutional neural network and GPU acceleration, has been demonstrated for real-time applications like professional table tennis matches. AI
IMPACT This system could enhance sports analytics and broadcast technologies by providing precise real-time ball spin data.
RANK_REASON The cluster contains a research paper detailing a novel system for spin estimation in ball games.
- arXiv
- baseball
- convolutional neural network
- Event camera
- golf
- graphics processing unit
- Naoya Takahashi
- pan/tilt galvanometer mirrors
- table tennis
- telephoto lens
- tennis
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
