摘要
提出了一种基于改进YOLOv5s的塔机作业中坠落危险区域识别方法。通过在目标检测网络中嵌入通道注意力模块、添加小目标检测层、更改边框损失函数改进优化YOLOv5s模型,利用改进YOLOv5s模型识别吊钩并确定坠物落点范围。结果表明,改进YOLOv5s模型检测效果得到提升,且模型占用计算机资源小。
A tower crane dangerous area identification model based on improved YOLOv5s is proposed.The YOLOv5s model is improved and optimized by embedding the channel attention module in the network,adding a scale detection layer and changing the bounding box regression loss function.The improved YOLOv5s model is used to identify the hook and determine the range of falling objects.The results show that the performance of the improved YOLOv5s model is improved,and the model is light weight.
作者
路学莹
孙啸
田正宏
王铫
LU Xueying;SUN Xiao;TIAN Zhenghong;WANG Yao(College of Water Conservancy and Hydropower,Hohai University,Nanjing 210024,Jiangsu,China;State Key Laboratory of Hydrology-Water Resources and Hydraulic Engineering,Hohai University,Nanjing 210024,Jiangsu,China)
出处
《水力发电》
CAS
2023年第2期68-77,共10页
Water Power
基金
国家自然科学基金青年基金资助项目(51909072)。
