This paper presents a hierarchical framework for real-time multi-person tracking for pose estimation. Current state-of-the-art systems face significant challenges with occlusion handling, identity preservation across frames, and accurate attribution of actions to specific individuals in multi-person contexts. We propose a multi-model framework within a hierarchical processing architecture to resolve this. The key contribution lies in decomposing the complex multi-person estimation problem into more tractable person-specific sub-problems, enabling high-precision pose estimation for multiple individuals while maintaining computational efficiency. Our framework facilitates temporally coherent pose tracking by preserving identity consistency across frames. Our contribution advances multi-person analysis through the effective integration of the model that overcomes limitations in pose estimation.