Single-target detection of Oncomelania hupensis based on improved YOLOv5s

Abstract

To address the issues of low detection accuracy and poor effect caused by small Oncomelania hupensis data samples and small target sizes. This article proposes the O. hupensis snails detection algorithm, the YOLOv5s-ECA-vfnet based on improved YOLOv5s, by using YOLOv5s as the basic target detection model and optimizing the loss function to improve target learning ability for specific regions. The experimental findings show that the snail detection method of the YOLOv5s-ECA-vfnet, the precision (P), the recall (R) and the mean Average Precision (mAP) of the algorithm are improved by 1.3%, 1.26%, and 0.87%, respectively. It shows that this algorithm has a good effect on snail detection. The algorithm is capable of accurately and rapidly identifying O. hupensis snails on different conditions of lighting, sizes, and densities, and further providing a new technology for precise and intelligent investigation of O. hupensiss snails for schistosomiasis prevention institutions.