Human Pose Estimation Based on an Improved YOLOv8n-Pose Algorithm

To address the limitations of traditional human pose estimation models in detection accuracy and robustness, this paper proposes an improved YOLOv8n-Pose algorithm. The proposed method enhances feature extraction and keypoint detection by introducing a parameter-free attention mechanism and a lightweight decoupled detection head, thereby improving accuracy and inference speed. Experimental results show that the improved algorithm achieves higher detection accuracy on the COCO dataset subset. While maintaining the efficiency of the original YOLOv8n-Pose model, the proposed approach demonstrates strong adaptability to complex scenarios and effectively balances the trade-off between speed and accuracy.