软件导刊2025，Vol.24Issue(6)：41-48,8.DOI:10.11907/rjdk.241413
一种基于改进型YOLOv5s的结直肠息肉检测算法QB-YOLO
QB-YOLO:Colorectal Polyp Detection Algorithm Based on Improved YOLOv5s
摘要
Abstract
In the field of medical imaging,early detection of colorectal polyps is crucial for preventing diseases such as colorectal cancer.In practical medical operations,the accuracy of automated detection of colorectal polyps is limited by many special conditions.Therefore,a colon polyp detection model QB-YOLO based on improved YOLOv5s is proposed.Firstly,a local context information enhancement module(CAM module)is introduced into the original backbone network to replace the spatial pyramid pooling SPPF module in the original model,in order to enhance the model's target attention to colorectal polyps;Secondly,adding a Large Kernel Separation Convolutional Attention Module(LSKA module)to the backbone network enhances the model's ability to capture local details in colorectal polyp images;Finally,Soft NMS with soft non maximum suppression is introduced into the model to address the possibility of dense distribution of some colorectal polyps,making the model more efficient in handling overlapping and dense targets.The experiment shows that the accuracy,recall,and average precision of the improved model have increased by 4.1%,8.5%,and 3.9%compared to the original model.关键词
YOLOv5s/结直肠息肉/目标检测/局部上下文信息增强/大核分离卷积注意力/软非极大值抑制Key words
YOLOv5s/colorectal polyps/object detection/local context augmentation/large separable kernel attention/Soft-NMS分类
信息技术与安全科学引用本文复制引用
张子健,徐建宇,杨欢..一种基于改进型YOLOv5s的结直肠息肉检测算法QB-YOLO[J].软件导刊,2025,24(6):41-48,8.基金项目
广东省基础与应用基础研究基金项目(2020A1515110783) （2020A1515110783）
