GTATrack: Winner Solution to SoccerTrack 2025 with Deep-EIoU and Global Tracklet Association

TL;DR

GTATrack is a hierarchical multi-object tracking framework that combines Deep-EIoU and Global Tracklet Association, achieving state-of-the-art accuracy in soccer tracking with fisheye cameras and addressing challenges like occlusions and scale variation.

Contribution

The paper introduces GTATrack, a novel hierarchical tracking framework with Deep-EIoU and GTA components, specifically designed for challenging sports scenarios with fisheye distortion.

Findings

Achieved first place in SoccerTrack Challenge 2025.

Attained a HOTA score of 0.60, outperforming previous methods.

Significantly reduced false positives to 982.

Abstract

Multi-object tracking (MOT) in sports is highly challenging due to irregular player motion, uniform appearances, and frequent occlusions. These difficulties are further exacerbated by the geometric distortion and extreme scale variation introduced by static fisheye cameras. In this work, we present GTATrack, a hierarchical tracking framework that win first place in the SoccerTrack Challenge 2025. GTATrack integrates two core components: Deep Expansion IoU (Deep-EIoU) for motion-agnostic online association and Global Tracklet Association (GTA) for trajectory-level refinement. This two-stage design enables both robust short-term matching and long-term identity consistency. Additionally, a pseudo-labeling strategy is used to boost detector recall on small and distorted targets. The synergy between local association and global reasoning effectively addresses identity switches, occlusions, and tracking fragmentation. Our method achieved a winning HOTA score of 0.60 and significantly reduced false positives to 982, demonstrating state-of-the-art accuracy in fisheye-based soccer tracking. Our code is available at https://github.com/ron941/GTATrack-STC2025.