重庆科技大学学报（自然科学版）2024，Vol.26Issue(6)：65-71,7.DOI:10.19406/j.issn.2097-4531.2024.06.010
改进YOLOv8n的齿面缺陷检测研究
Detection of Tooth Surface Defects Based on Improved YOLOv8n
摘要
Abstract
In order to solve the problems of low detection efficiency and poor detection consistency in manual quality inspection of rough gears,a YOLO-CHD algorithm model based on improved YOLOv8n is proposed.Firstly,the C2f layer is added after the initial convolution of the backbone network to reduce the loss of small target feature de-tails by the subsequent convolutional layer.Secondly,the shallow large-size feature map is further fused in the fea-ture fusion network,and the small target is detected on the higher-resolution feature map with ASFF-4H to im-prove the detection accuracy.Finally,the C2f of the deep feature map in the feature fusion network is replaced by Dual-C2f to further improve the detection accuracy of the model.The experimental results show that the average accuracy of the improved model reaches 73.3%,which is 3.3 percentage points higher than that of the original model,and the inference speed is 42 frames/s,which basically meets the needs of rough machining quality inspec-tion in the tooth surface defect process.关键词
缺陷检测/粗加工齿轮/特征增强/特征融合/YOLOv8n算法Key words
defect detection/roughing gears/feature enhancement/feature fusion/YOLOv8n algorithm分类
信息技术与安全科学引用本文复制引用
马彬洋,彭龑..改进YOLOv8n的齿面缺陷检测研究[J].重庆科技大学学报（自然科学版）,2024,26(6):65-71,7.基金项目
自贡市科技局科技计划资助项目"云端协同工作的虚拟机在线迁移性能研究"(2018GYCX33) （2018GYCX33）
