TOTNet: Occlusion-Aware Temporal Tracking for Robust Ball Detection in Sports Videos

2025-08-13

AI Summary

Insufficient robustness in ball tracking under occlusion in sports videos severely hampers event detection and referee assistance. To address this, we propose TOMNet—a ball detection and tracking framework specifically designed for highly occluded scenarios. TOMNet innovatively integrates 3D convolutional networks to model spatiotemporal dynamics, a visibility-weighted loss function that explicitly encodes occlusion states, and a targeted occlusion-aware data augmentation strategy. Furthermore, we introduce TTA, the first publicly available table-tennis benchmark dataset featuring high occlusion rates. Extensive experiments across four public datasets demonstrate that TOMNet significantly improves tracking accuracy: RMSE decreases from 37.30 to 7.19 pixels, and localization accuracy on fully occluded frames rises from 0.63 to 0.80. TOMNet consistently outperforms state-of-the-art methods across all evaluated metrics.

Technology Category

• Computer Vision: Motion & Tracking
• Intelligent Robots: State Estimation
• Reasoning under Uncertainty: Other Foundations of Reasoning under Uncertainty

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Robust ball tracking under occlusion remains a key challenge in sports video analysis, affecting tasks like event detection and officiating. We present TOTNet, a Temporal Occlusion Tracking Network that leverages 3D convolutions, visibility-weighted loss, and occlusion augmentation to improve performance under partial and full occlusions. Developed in collaboration with Paralympics Australia, TOTNet is designed for real-world sports analytics. We introduce TTA, a new occlusion-rich table tennis dataset collected from professional-level Paralympic matches, comprising 9,159 samples with 1,996 occlusion cases. Evaluated on four datasets across tennis, badminton, and table tennis, TOTNet significantly outperforms prior state-of-the-art methods, reducing RMSE from 37.30 to 7.19 and improving accuracy on fully occluded frames from 0.63 to 0.80. These results demonstrate TOTNets effectiveness for offline sports analytics in fast-paced scenarios. Code and data access:href{https://github.com/AugustRushG/TOTNet}{AugustRushG/TOTNet}.

Problem

• Robust ball tracking under occlusion in sports videos
• Improving detection accuracy during partial and full occlusions
• Addressing challenges in fast-paced sports analytics scenarios

Innovation

• Uses 3D convolutions for temporal tracking
• Employs visibility-weighted loss for occlusion handling
• Applies occlusion augmentation to improve robustness