Research on fall detection algorithm based on improved YOLOv8

Aiming at the traditional fall detection algorithms that suffer from low detection accuracy and poor real-time performance in dealing with small targets and multi-scale features in complex scenes, this study proposes a fall detection algorithm based on an improved YOLOv8 model. First, the SCConv module is utilized to combine with the C2f module in the YOLOv8 backbone network to construct the C2fSC module, which enhances the feature extraction capability of the model while reducing the computational complexity. Second, the SPPF module is integrated with the BAM attention mechanism to weaken the influence of background interference. Finally, WIoU is used to replace the original loss function to optimize the positioning accuracy of the bounding box and enhance the overall performance of the model. The experimental results show that compared with the original YOLOv8 model, the improved model improves 4.9%, 4.2%, and 4.4% in P, R, and mAP metrics, respectively, and the number of parameters and FLOPs are only 2.8M and 7.7G, which proves that the improved model has excellent robustness and effectiveness.