Problem
Research questions and friction points this paper is trying to address.
Long-Term Multi-Object Tracking
Sports Video Analytics
Occlusion
Camera Motion
Identity Preservation
Training-Free Long-Term Multi-Object Tracking for Sports Video Analytics
🤖 AI Summary
This study addresses the challenges of occlusion, camera motion, and target re-identification in long-term multi-object tracking for sports videos by proposing McByte++, a training-free framework. The method establishes a unified pipeline integrating lightweight mask propagation, conditional camera motion compensation, online re-identification, and offline global association. By replacing heavy segmentation modules with lightweight components and optimizing motion modeling, it enhances tracking robustness without detector retraining. Experimental results demonstrate that McByte++ improves HOTA and IDF1 scores by 3.0 and 6.1 points, respectively, while achieving an order-of-magnitude increase in inference speed. These findings indicate significant advancements in both tracking efficiency and identity preservation capabilities for complex sports scenarios.
📝 Abstract
Long-term multi-object tracking in sports remains challenging due to frequent occlusions, rapid camera motion, and repeated player reappearances. We introduce McByte++, a training-free tracking-by-detection framework that integrates lightweight mask propagation, conditional camera motion compensation, and online re-identification within a unified pipeline. Compared to its predecessor, McByte++ substantially improves runtime efficiency while enhancing identity preservation. On SoccerNet-tracking and SportsMOT benchmarks, McByte++ achieves up to +3.0 HOTA and +6.1 IDF1 improvements over the original McByte in the online setting, with further gains when combined with offline global association. Replacing heavy segmentation components and optimizing motion modeling yields up to an order-of-magnitude speed increase. All results are obtained without detector retraining or dataset-specific tuning. Code will be made available at https://github.com/tstanczyk95/McBytePlusPlus.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Training-Free Tracking
Lightweight Mask Propagation
Conditional Camera Motion Compensation
Online Re-identification
Runtime Efficiency
🔎 Similar Papers
No similar papers found.
