基于改进YOLOv5的变电站屋面工程缺陷检测算法研究OA
Research on detection algorithm for substation roof engineering defects using an improved YOLOv5 model
针对变电站建筑物屋面工程缺陷检测效率较低及检测精确度较差的问题,提出一种基于改进YOLOv5(you only look once version 5)的变电站屋面工程缺陷检测算法.首先,对图像进行预处理,减轻外界噪声给检测效果带来的影响.其次,在网络骨干中引入改进自注意力机制,提高计算效率,用多头自注意力层替换YOLOv5 网络骨干末端的卷积层,使网络能够更好地捕捉全局关联信息.最后,在检测部分增加跨层加权级联结构,将浅层中缺陷的边缘信息、轮廓…查看全部>>
This paper proposes a detection algorithm based on an improved you only look once version 5(YOLOv5)model to address the issues of low efficiency and poor accuracy in defect detection of substation roof constructions.Firstly,the algorithm preprocesses images to reduce the impact of external noise on the detection outcomes.Secondly,it intro-duces an improved self-attention mechanism into the network's backbone.It replaces the traditional convolutional layers a…查看全部>>
张晓晨;徐波
国网宁夏电力有限公司建设分公司,宁夏 银川 750004国网宁夏电力有限公司建设分公司,宁夏 银川 750004
动力与电气工程
屋面工程缺陷检测深度学习YOLOv5自注意力跨层加权级联
roof engineering defect detectiondeep learningYOLOv5self-attentioncross-layer weighted cascading
《宁夏电力》 2024 (3)
70-76,7
国网宁夏建设分公司2023年群众性科技创新项目(5229JS230002)
评论