BlurBall: Joint Ball and Motion Blur Estimation for Table Tennis Ball Tracking

2025-09-22

AI Summary

High-speed motion of table tennis balls induces severe motion blur—appearing as streaks—which introduces localization bias when conventional front-edge annotations are used and discards critical velocity cues. Method: This paper proposes a novel annotation paradigm centered on the blur streak’s geometric center, explicitly modeling motion attributes including blur direction and length. We design an end-to-end multi-frame network incorporating Squeeze-and-Excitation attention to jointly predict ball centroid coordinates and motion-related features. Contributions: (1) We release the first table tennis detection dataset with center-aligned motion-blur annotations; (2) We empirically demonstrate that explicit modeling of blur features significantly enhances robustness for small-object detection and enables accurate velocity estimation; (3) Our approach achieves state-of-the-art detection performance across multiple mainstream architectures and substantially improves both accuracy and stability of trajectory prediction.

Technology Category

• Computer Vision: Motion & Tracking
• Machine Learning: Multi-instance/Multi-view Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Motion blur reduces the clarity of fast-moving objects, posing challenges for detection systems, especially in racket sports, where balls often appear as streaks rather than distinct points. Existing labeling conventions mark the ball at the leading edge of the blur, introducing asymmetry and ignoring valuable motion cues correlated with velocity. This paper introduces a new labeling strategy that places the ball at the center of the blur streak and explicitly annotates blur attributes. Using this convention, we release a new table tennis ball detection dataset. We demonstrate that this labeling approach consistently enhances detection performance across various models. Furthermore, we introduce BlurBall, a model that jointly estimates ball position and motion blur attributes. By incorporating attention mechanisms such as Squeeze-and-Excitation over multi-frame inputs, we achieve state-of-the-art results in ball detection. Leveraging blur not only improves detection accuracy but also enables more reliable trajectory prediction, benefiting real-time sports analytics.

Problem

• Motion blur reduces clarity of fast balls in racket sports
• Existing methods inaccurately label ball position ignoring motion cues
• Need joint estimation of ball position and blur attributes

Innovation

• Centers ball labels within blur streaks
• Jointly estimates position and blur attributes
• Uses attention mechanisms on multi-frame inputs