基于YOLOv8算法的常用手势识别

Common gesture recognition based on YOLOv8 algorithm

手势控制医疗设备是一种新型的人机交互方式,其通过非接触式手势识别技术,为医护人员提供更加便捷、高效、卫生的操作方式。本文首先介绍了基于计算机视觉手势识别的概念,讨论了手势识别的分类等关键技术。继而提出了基于计算机视觉的手势识别YOLO算法,旨在实现对医疗设备的便捷操作。实验验证了YOLOv8算法在手势识别方面的可行性,并与YOLOv5版本进行了比较。实验结果显示,尽管YOLOv8的检测速度稍慢,但在不同IOU阈值下的平均精度值更高。通过使用YOLOv8模型训练的常用控制手势识别模型在测试集中展现出出色的拟合性能,能够实现对图片、视频和摄像头的实时检测与识别。

Gesture-controlled medical devices are a novel form of human-computer interaction that utilizes non-contact gesture recognition technology to provide healthcare professionals with a more convenient,efficient,and hygienic mode of operation.This paper proposed a computer vision-based gesture recognition YOLO algorithm after conducting classification and key technology research on gesture recognition.The feasibility of the YOLOv8 algorithm for gesture recognition operations was verified through experiments and compared with the YOLOv5 version.The experiments demonstrated that the designed gesture recognition algorithm model,trained using the YOLOv8 model,was capable of recognizing commonly used control gestures for medical devices.Although YOLOv8 exhibits slower detection speeded compared to YOLOv5,it achieved better average precision values at different IOU thresholds.The trained model showed excellent performance in fitting the test set,enabling real-time detection and recognition of gestures in images,videos,and from cameras.