Research on Improvement of sports Posture Key Point Detection Algorithm based on EDI-YOLOv8 Algorithm

In response to the limitations of existing sports posture recognition algorithms in terms of accuracy and efficiency, this paper proposes an efficient detection algorithm, EDI-YOLOv8, based on an improved YOLOv8-pose. First, an ECA channel attention mechanism is added to the Neck section to capture long-range dependencies, effectively integrating multi-scale information. The DCNv2 module is used to reconstruct C2f, enabling the model to effectively capture fine-grained target details and adaptively adjust the sampling positions of the convolution kernels. The Inner-SIoU is employed as a new localization regression loss function to enhance the learning ability for small target samples and accelerate the convergence of the bounding box regression. Experimental results show that the detection accuracy of the proposed algorithm reaches 74.5%, which is 3.3% higher than the original model. Compared to the original YOLOv8-pose model, the improved model not only achieves higher detection accuracy but also shows significant improvement in the recognition of small targets. Therefore, the improved network can be applied more accurately and effectively to human sports action posture detection.