This chapter comprehensively reviews the evolution of single object tracking algorithms, ranging from traditional methods to cutting-edge vision-language model (LVM)-based trackers. Traditional tracking methods emphasize foundational components such as motion modeling, feature representation, and model updating, providing the groundwork for the field. Correlation filter-based trackers introduce computational efficiency and adaptability, while Siamese Neural Networks (SNNs) revolutionize tracking with feature-matching paradigms, exemplified by SiamFC and SiamRPN. Transformer-based trackers like TransT and SwinTrack leverage attention mechanisms to enhance robustness and contextual understanding. Lastly, LVM-based trackers, such as SAM-Track and TrackGPT, integrate multimodal and instruction-driven capabilities, representing a paradigm shift toward dynamic and semantic tracking. The chapter illustrates the continuous progression of SOT, offering insights into past innovations and future directions.