The field of sports video understanding is moving towards more nuanced and detailed analysis of fast-paced and complex sports scenarios. Researchers are developing innovative methods to improve the accuracy and reliability of video understanding models, particularly in domains where existing approaches struggle, such as tennis and table tennis. A key direction is the incorporation of multimodal and temporal information to better capture the dynamics of sports events. Another area of focus is the development of specialized frameworks and benchmarks that can handle the unique challenges of sports video understanding, such as motion blur and rapid object movement. Noteworthy papers include: AdaSports-Traj, which proposes an adaptive trajectory modeling framework for multi-agent sports scenarios, and BlurBall, which introduces a new labeling strategy and model for joint ball and motion blur estimation in table tennis.