深度卷积神经网络支持下的遥感影像飞机检测
|
摘要
针对YOLOv3算法对小目标检测较差及出现较多漏检的问题,本文提出了一种优化的YOLOv3算法。首先使用K-means算法计算出与数据集相适用的锚框;其次将扩张卷积引入到YOLOv3网络,用来增强网络高层的感受野,改善小目标的检测效果;然后使用深度可分离卷积取代YOLOv3网络残差模块中的普通卷积,可减少计算量,从而得到一种新型卷积神经网络结构;最后在数据集上进行对比试验。结果表明,优化的YOLOv3算法能够检测出更多目标,降低漏检率,相比于YOLOv3算法,其召回率提高11.86%,F1-score提高2.99%。
Aiming at the problem that YOLOv3( You look only once) algorithm has poor detection of small targets and many missed detection,this paper proposes an optimized YOLOv3 algorithm. Firstly,K-means algorithm is used to calculate the anchor frame suitable for the data set in this paper. Then,extended convolution is introduced into YOLOv3 network to enhance the high-level perception field of the network,improve the detection effect of small targets,and secondly,depthwise separable convolution is used. It replaces the ordinary convolution in YOLOv3 network residual module,reduces the calculation parameters,and thus obtains a new convolution neural network structure. Then the comparative experiments are carried out on the data set in this paper. The experimental results show that the optimized YOLOv3 algorithm can detect more targets and reduce the missed detection rate. Compared with YOLOv3 algorithm,its recall rate is increased by 11.86% and F1-score by 2.99%.
Aiming at the problem that YOLOv3( You look only once) algorithm has poor detection of small targets and many missed detection,this paper proposes an optimized YOLOv3 algorithm. Firstly,K-means algorithm is used to calculate the anchor frame suitable for the data set in this paper. Then,extended convolution is introduced into YOLOv3 network to enhance the high-level perception field of the network,improve the detection effect of small targets,and secondly,depthwise separable convolution is used. It replaces the ordinary convolution in YOLOv3 network residual module,reduces the calculation parameters,and thus obtains a new convolution neural network structure. Then the comparative experiments are carried out on the data set in this paper. The experimental results show that the optimized YOLOv3 algorithm can detect more targets and reduce the missed detection rate. Compared with YOLOv3 algorithm,its recall rate is increased by 11.86% and F1-score by 2.99%.
引文
[1]DING P,ZHANG Y,DENG W J,et al.A light and faster regional convolutional neural network for object detection in optical remote sensing images[J].ISPRSJournal of Photogrammetry&Remote Sensing,2018,141:208-128.
[2]蔡栋,陈焱明,魏巍.基于骨架特征的多光谱遥感影像飞机目标识别[J].测绘通报,2014(2):50-54.
[3]马琦,马蔚鹏,刘彦,等.基于支持向量机的图像飞机目标自动识别算法研究[J].计算机测量与控制,2014,22(9):2851-2852.
[4]姚远,姜志国,张浩鹏.基于层次化分类器的遥感图像飞机目标检测[J].航天返回与遥感,2014,35(5):88-94.
[5]CHAN T H,JIA K,GAO S,et al.PCANet:a simple deep learning baseline for image classification?[J].IEEE Transactions on Image Processing,2015,24(12):5017-5032.
[6]COLLOBERT R,WESTON J.A unified architecture for natural language processing:deep neural networks with multitask learning[C]∥International Conference on Machine Learning.Helsinki:ICML,2008:160-167.
[7]BENGIO Y.Learning deep architectures for AI[M].[S.l.]:Now Publisher Inc.,2009.
[8]MOHAMED A R,SAINATH T N,DAHL G,et al.Deep belief networks using discriminative features for phone recognition[C]∥IEEE International Conference on Acoustics,Speech and Signal Processing.Prague:IEEE,2011:5060-5063.
[9]ZHANG F,DU B,ZHANG L.Scene classification via a gradient boosting random convolutional network framework[J].IEEE Transactions on Geoscience&Remote Sensing,2016,54(3):1793-1802.
[10]MA X,GENG J,WANG H.Hyperspectral image classification via contextual deep learning[J].Eurasip Journal on Image&Video Processing,2015(1):20.
[11]HU W,HUANG Y,WEI L,et al.Deep convolutional neural networks for hyperspectral image classification[J].Journal of Sensors,2015(2):1-12.
[12]殷文斌,王成波,袁翠,等.一种飞机目标的遥感识别方法[J].测绘通报,2017(3):37-40.
[13]邓志鹏,孙浩,雷琳,等.基于多尺度形变特征卷积网络的高分辨率遥感影像目标检测[J].测绘学报,2018,47(9):1216-1227.
[14]张志远.基于深度学习的光学遥感图像飞机检测[D].厦门:厦门大学,2016.
[15]CHEN C,LIU M Y,TUZEL O,et al.R-CNN for small object detection[C]∥Asian Conference on Computer Vision.Taipei:IEEE,2016:214-230.
[16]HE K,ZHANG X,REN S,et al.Spatial pyramid pooling in deep convolutional networks for visual recognition[J].IEEE Transactions on Pattern Analysis&Machine Intelligence,2015,37(9):1904-1916.
[17]GIRSHICK R.Fast R-CNN[C]∥Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Santiago:IEEE,2015.
[18]REN S,HE K,GIRSHICK R,et al.Faster R-CNN:towards real-time object detection with region proposal networks[J].IEEE Trans Pattern Anal Mach Intell,2015,39(6):1137-1149.
[19]REDMON J,DIVVALA S,GIRSHICK R,et al.You only look once:unified,real-time object detection[C]∥Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Las Vegas:IEEE,2016.
[20]LIU W,ANGUELOV D,ERHAN D,et al.SSD:single shot multibox detector[C]∥Proceedings of the IEEEConference on Computer Vision and Pattern Recognition.Las Vegas:IEEE,2016.
[21]郑志强,刘妍妍,潘长城,等.改进YOLO V3遥感图像飞机识别应用[EB/OL].(2018-01-04)[2019-02-01].http:∥kns.cnki.net/kcms/detail/41.1227.TN.20180823.1037.002.html.
[22]刘学平,李玙乾,刘励,等.嵌入SENet结构的改进YOLOV3目标识别算法[EB/OL].(2017-12-19)[2018-12-13].https:∥doi.org/10.19678/j.issn.1000-3428.0052861.
[23]CHEN L C,PAPANDREOU G,SCHROFF F,et al.Rethinking atrous convolution for semantic image segmentation[C]∥Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Puerto Rico:IEEE,2017.
[24]WANG P,CHEN P,YUAN Y,et al.Understanding convolution for semantic segmentation[C]∥IEEEWinter Conference on Applications of Computer Vision.[S.l.]:IEEE,2018.
[25]CHOLLET F.Xception:deep learning with depthwise separable convolutions[C]∥Proceedings of the IEEEConference on Computer Vision and Pattern Recognition.Puerto Rico:IEEE,2017.
[26]LI Z,PENG C,YU G,et al.Det Net:a backbone network for object detection[C]∥Proceedings of the IEEEConference on Computer Vision and Pattern Recognition.Salt Lake City:IEEE,2018.
[2]蔡栋,陈焱明,魏巍.基于骨架特征的多光谱遥感影像飞机目标识别[J].测绘通报,2014(2):50-54.
[3]马琦,马蔚鹏,刘彦,等.基于支持向量机的图像飞机目标自动识别算法研究[J].计算机测量与控制,2014,22(9):2851-2852.
[4]姚远,姜志国,张浩鹏.基于层次化分类器的遥感图像飞机目标检测[J].航天返回与遥感,2014,35(5):88-94.
[5]CHAN T H,JIA K,GAO S,et al.PCANet:a simple deep learning baseline for image classification?[J].IEEE Transactions on Image Processing,2015,24(12):5017-5032.
[6]COLLOBERT R,WESTON J.A unified architecture for natural language processing:deep neural networks with multitask learning[C]∥International Conference on Machine Learning.Helsinki:ICML,2008:160-167.
[7]BENGIO Y.Learning deep architectures for AI[M].[S.l.]:Now Publisher Inc.,2009.
[8]MOHAMED A R,SAINATH T N,DAHL G,et al.Deep belief networks using discriminative features for phone recognition[C]∥IEEE International Conference on Acoustics,Speech and Signal Processing.Prague:IEEE,2011:5060-5063.
[9]ZHANG F,DU B,ZHANG L.Scene classification via a gradient boosting random convolutional network framework[J].IEEE Transactions on Geoscience&Remote Sensing,2016,54(3):1793-1802.
[10]MA X,GENG J,WANG H.Hyperspectral image classification via contextual deep learning[J].Eurasip Journal on Image&Video Processing,2015(1):20.
[11]HU W,HUANG Y,WEI L,et al.Deep convolutional neural networks for hyperspectral image classification[J].Journal of Sensors,2015(2):1-12.
[12]殷文斌,王成波,袁翠,等.一种飞机目标的遥感识别方法[J].测绘通报,2017(3):37-40.
[13]邓志鹏,孙浩,雷琳,等.基于多尺度形变特征卷积网络的高分辨率遥感影像目标检测[J].测绘学报,2018,47(9):1216-1227.
[14]张志远.基于深度学习的光学遥感图像飞机检测[D].厦门:厦门大学,2016.
[15]CHEN C,LIU M Y,TUZEL O,et al.R-CNN for small object detection[C]∥Asian Conference on Computer Vision.Taipei:IEEE,2016:214-230.
[16]HE K,ZHANG X,REN S,et al.Spatial pyramid pooling in deep convolutional networks for visual recognition[J].IEEE Transactions on Pattern Analysis&Machine Intelligence,2015,37(9):1904-1916.
[17]GIRSHICK R.Fast R-CNN[C]∥Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Santiago:IEEE,2015.
[18]REN S,HE K,GIRSHICK R,et al.Faster R-CNN:towards real-time object detection with region proposal networks[J].IEEE Trans Pattern Anal Mach Intell,2015,39(6):1137-1149.
[19]REDMON J,DIVVALA S,GIRSHICK R,et al.You only look once:unified,real-time object detection[C]∥Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Las Vegas:IEEE,2016.
[20]LIU W,ANGUELOV D,ERHAN D,et al.SSD:single shot multibox detector[C]∥Proceedings of the IEEEConference on Computer Vision and Pattern Recognition.Las Vegas:IEEE,2016.
[21]郑志强,刘妍妍,潘长城,等.改进YOLO V3遥感图像飞机识别应用[EB/OL].(2018-01-04)[2019-02-01].http:∥kns.cnki.net/kcms/detail/41.1227.TN.20180823.1037.002.html.
[22]刘学平,李玙乾,刘励,等.嵌入SENet结构的改进YOLOV3目标识别算法[EB/OL].(2017-12-19)[2018-12-13].https:∥doi.org/10.19678/j.issn.1000-3428.0052861.
[23]CHEN L C,PAPANDREOU G,SCHROFF F,et al.Rethinking atrous convolution for semantic image segmentation[C]∥Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Puerto Rico:IEEE,2017.
[24]WANG P,CHEN P,YUAN Y,et al.Understanding convolution for semantic segmentation[C]∥IEEEWinter Conference on Applications of Computer Vision.[S.l.]:IEEE,2018.
[25]CHOLLET F.Xception:deep learning with depthwise separable convolutions[C]∥Proceedings of the IEEEConference on Computer Vision and Pattern Recognition.Puerto Rico:IEEE,2017.
[26]LI Z,PENG C,YU G,et al.Det Net:a backbone network for object detection[C]∥Proceedings of the IEEEConference on Computer Vision and Pattern Recognition.Salt Lake City:IEEE,2018.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |