This paper introduces a novel semi-supervised on-line boosting method for robust tracking, addressing the key issue of drifting in online adaptation. The method combines a priori knowledge from labeled data with online updates from unlabeled samples, allowing the tracker to adapt to appearance changes while maintaining stability. The main contribution is an on-line formulation of semi-supervised boosting, which is essential for real-time tracking applications. Experiments demonstrate that the proposed SemiBoost tracker outperforms other on-line tracking methods in challenging scenarios, effectively handling various object appearances and occlusions. The approach is parameter-free and easy to implement, making it a robust and adaptive solution for tracking tasks.This paper introduces a novel semi-supervised on-line boosting method for robust tracking, addressing the key issue of drifting in online adaptation. The method combines a priori knowledge from labeled data with online updates from unlabeled samples, allowing the tracker to adapt to appearance changes while maintaining stability. The main contribution is an on-line formulation of semi-supervised boosting, which is essential for real-time tracking applications. Experiments demonstrate that the proposed SemiBoost tracker outperforms other on-line tracking methods in challenging scenarios, effectively handling various object appearances and occlusions. The approach is parameter-free and easy to implement, making it a robust and adaptive solution for tracking tasks.