基于YOLOv5的新冠肺炎疾病的检测

Detection of COVID-19 Disease Based on YOLOv5

摘要: 新冠肺炎是一种流行性的传染性疾病，世界各国都在爆发。随着病例数量的逐渐增多，医生的压力也逐渐增大，人工智能的日渐深入研究与不断发展，国内外专家学者也都致力于将计算机辅助检测与诊断应用于医学影像领域的研究，并且在医学之中应用范围逐渐扩大，为了缓解医生的对于确诊病例的诊断，本文采用YOLOv5对病人的CT图像进行标注，以辅助医生诊断。通过实验证实，采用YOLOv5算法可以有效地对新冠肺炎疾病与正常肺部进行判断，有效精准地预测了实验结果，经过三十轮的数据检测，预测的mAP_0.5达到了99.5%，mAP_0.5:0.95达到99.1%。利用YOLOv5计算机辅助医学检测将有助于提升对新冠肺炎疾病准确快速的检测。

Abstract: COVID-19 is an epidemic infectious disease that is breaking out in countries all over the world. With the gradual increase of the number of cases, the pressure of doctors is increasing, and the deepening research and continuous development of artificial intelligence, experts and scholars at home and abroad are also committed to applying computer-aided detection and diagnosis to the research of medical imaging, and the application scope in medicine is gradually expanding. In order to alleviate the doctors’ diagnosis of confirmed cases, this paper uses YOLOv5 to mark the CT images of patients to assist doctors in diagnosis. The experiment proved that the YOLOv5 algorithm can effectively judge the COVID-19 disease and normal lungs, and effectively and accurately predict the experimental results. After 30 rounds of data detection, the predicted mAP_0.5 reached 99.5%, and mAP_0.05:0.95 reached 99.1%. The use of YOLOv5 computer-aided medical testing will help to improve the accurate and rapid detection of COVID-19.