基于特征金字塔网络与树莓派的护理床智能控制方法研究

Intelligent Control Method of Nursing Bed Combining Feature Pyramid Network and Raspberry Pi

摘要:

为解决传统护理床中存在的效率低、操作复杂等问题,研究通过特征金字塔进行手势检测,并引入通道注意力与Transformer注意力对特征金字塔进行优化,并在树莓派的基础上设计了一个护理床控制系统,然后将优化后的特征金字塔应用于其中,从而设计出一种结合特征金字塔网络与树莓派的护理床智能控制系统。结果显示,改进模型在COCO数据集上的准确率可达95%。在角度测试误差中,改进模型的最小误差率为1.17%,证明了其精度较高。在识别率与平均测试时间中,改进模型的识别率在不同指令中的最高可达98.7%,平均测试时间为0.18s,证明了其有效性,并进一步证明了其准确性。基于该控制方法的智能护理床能够有效提高老年人的护理质量和舒适度,为智能护理床的进一步发展提供了新的方向。

Abstract:

In order to solve the problems of low efficiency and complex operation in traditional nursing beds, this study investigates gesture detection through feature pyramids, and introduces channel attention and Transformer attention to optimize the feature pyramids. Based on Raspberry Pi, a nursing bed control system is designed, and the optimized feature pyramids are applied to it, thus designing an intelligent control system for nursing beds that combines feature pyramid networks with Raspberry Pi. The results show that the accuracy of the improved model on the COCO dataset can reach 95%. In the angle testing error, the minimum error rate of the improved model is 1.17%, which proves its high accuracy. In terms of recognition rate and average testing time, the highest recognition rate of the improved model in different instructions can reach 98.7%, and the average testing time is 0.18 seconds, which proves its effectiveness and further proves its accuracy. The intelligent nursing bed based on this control method can effectively improve the nursing quality and comfort of the elderly, providing a new direction for the further development of intelligent nursing beds.