摘要
电力巡检是保障电力稳定输送的重要因素.随着无人机巡检在电力巡检领域的广泛应用,产生大量的巡检图像需要人工进行检测.由于工作强度高、效率低且容易受人为因素影响使检测效果参差不齐.因此,基于计算机的目标检测算法对巡检数据进行处理十分必要.首先介绍基于传统特征提取目标检测和基于深度学习目标检测以及它们在电力巡检的应用;其次从目标检测应用在电力领域的趋势出发,具体探讨基于深度学习目标检测技术在电力巡检的应用前景;最后对巡检数据集的建立、网络模型轻量化和检测方法三个关键问题进行了重点阐述.
Power inspection is an important factor to ensure stable power transmission.With the widespread application of drone inspections in the field of power inspections,a large number of inspection images need to be manually inspected.Due to high work intensity,low efficiency and easy influence by human factors,the detection results are uneven.Therefore,it is necessary to process the inspection data based on the computer-based object detection algorithm.Firstly,target detection based on traditional feature extraction and deep learning as well as their application in power inspection are introduced.Secondly,beginning with the trend of target detection application in the power field,the application of target detection technology based on deep learning in power inspection prospects will be specifically discussed.Finally,three key issues,namely,the establishment of the inspection data set,the lightweight of the network model and the detection method are emphasized.
作者
张兆云
何冠锋
黄世鸿
张志
ZHANG Zhaoyun;HE Guanfeng;HUANG Shihong;ZHANG Zhi(College of Electronic Engineering and Intelligence,Dongguan University of Technology,Dongguan 523808,China)
出处
《湖北民族大学学报（自然科学版）》
CAS
2021年第3期305-314,共10页
Journal of Hubei Minzu University:Natural Science Edition
基金
广东省科技计划项目(2017A010104023).
关键词
目标检测
深度学习
电力巡检
图像识别
卷积网络
边缘计算
target detection
deep learning
power line inspection
image recognition
Convolution network
edge of computing
