Technical Report for ReID-SAM on SkiTB Visual Tracking Challenge 2025
Kunjun Li, Cheng-Yen Yang, Hsiang-Wei Huang, Jenq-Neng Hwang
TL;DR
This work tackles robust skier tracking in winter sports by introducing ReID-SAM, a pipeline that combines a SAMURAI-based tracker with an OSNet-driven Re-ID module and targeted post-processing. A mid-frame grounding step with GroundingDINO and a prompt-based reinitialization help correct identity switches, while YOLOv11-Kalman and STARK-based detections strengthen equipment tracking in single- and multi-skier scenarios. The approach delivers state-of-the-art results on the SkiTB dataset across alpine, jump, and freestyle skiing, demonstrating improved identity stability and comprehensive equipment coverage. The method offers practical value for automated skiing analytics, performance assessment, and robust multi-camera tracking in challenging winter environments.
Abstract
This report introduces ReID-SAM, a novel model developed for the SkiTB Challenge that addresses the complexities of tracking skier appearance. Our approach integrates the SAMURAI tracker with a person re-identification (Re-ID) module and advanced post-processing techniques to enhance accuracy in challenging skiing scenarios. We employ an OSNet-based Re-ID model to minimize identity switches and utilize YOLOv11 with Kalman filtering or STARK-based object detection for precise equipment tracking. When evaluated on the SkiTB dataset, ReID-SAM achieved a state-of-the-art F1-score of 0.870, surpassing existing methods across alpine, ski jumping, and freestyle skiing disciplines. These results demonstrate significant advancements in skier tracking accuracy and provide valuable insights for computer vision applications in winter sports.