电子测量与仪器学报

彭道刚,王永坤,周洋,戚尔江,高义民.基于改进 Faster R-CNN 的电厂雨排口污染物泄漏检测[J].电子测量与仪器学报,2022,36(2):40-48

基于改进 Faster R-CNN 的电厂雨排口污染物泄漏检测

Leakage detection of pollutants at rain drain outlet of powerplant based on improved Faster R-CNN

DOI：

中文关键词: 监控图像电厂雨排口废弃油污泄漏改进 Faster R-CNN

英文关键词:monitoring images power plant rainwater outlet waste oil leakage improved Faster R-CNN

基金项目:上海市“科技创新行动计划”高新技术领域项目（21511101800）、上海市科学技术委员会工程技术研究中心项目（14DZ2251100）资助

作者	单位
彭道刚	1. 上海电力大学自动化工程学院
王永坤	1. 上海电力大学自动化工程学院
周洋	2. 宝山钢铁股份有限公司电厂
戚尔江	1. 上海电力大学自动化工程学院
高义民	1. 上海电力大学自动化工程学院

Author	Institution
Peng Daogang	1. College of Automation Engineering,Shanghai University of Electric Power
Wang Yongkun	1. College of Automation Engineering,Shanghai University of Electric Power
Zhou Yang	2. Power Plant of Baoshan Iron & Steel Co. ,Ltd.
Qi Erjiang	1. College of Automation Engineering,Shanghai University of Electric Power
Gao Yimin	1. College of Automation Engineering,Shanghai University of Electric Power

摘要点击次数: 615

全文下载次数: 993

中文摘要:

针对监控图像中电厂雨排口出现的废弃油污泄漏问题,提出一种基于改进 Faster 区域卷积神经网络(Faster R-CNN)的电厂雨排口污染物泄漏检测算法。改进 Faster R-CNN 检测算法首先使用 ResNet-50 作为主干网络,在此基础上构建多尺度特征图金字塔结构(FPN),实现高层语义和低层语义之间的信息融合,提高了检测精度;其次采用 CIoU 损失和 DIoU-NMS 方法, 提高 Faster R-CNN 中边框回归的准确度;最后引入 Focal Loss 损失函数,解决了区域建议网络(RPN)生成的锚点冗余导致 RCNN 阶段出现正负样本不均衡问题。实验结果表明,此改进算法在真实样本中表现良好,平均准确率达到 90. 2%,与原 Faster R-CNN 算法相比较,准确率提高,误报率和漏报率明显下降,可有效应用于实际生产环境中。

英文摘要:

Aiming at the problem of waste oil leakage from power plant rainwater outlets in monitoring images, a pollutant leakage detection algorithm based on improved Faster R-CNN is proposed. The improved Faster R-CNN detection algorithm first uses ResNet-50 as the backbone network, and builds a multi-scale feature map pyramid structure ( FPN) on this basis to achieve information fusion between high-level semantics and low-level semantics, and improve detection accuracy; Secondly, the CIoU loss and DIoU-NMS methods are used to improve the accuracy of bounding box regression in Faster R-CNN; Finally, by introducing Focal Loss function, it solves the problem of unbalanced positive and negative samples in the R-CNN training stage caused by redundant anchor generated by RPN network. The experimental results show that the improved algorithm performs well in real samples, and the accuracy rate reaches 90. 2%. Compared with the original Faster R-CNN algorithm, the accuracy rate is improved, and the false positive rate and false negative rate are significantly reduced. It can be effectively used in the actual environment.

查看全文查看/发表评论下载PDF阅读器