面向航空影像下车辆目标的实时检测算法
|
摘要
为解决自然场景下的通用目标检测框架对航空影像下的小车辆目标检测性能不足的缺陷,提出一种专用于航空影像下的小车辆目标实时检测器,即轻量级尺度公平单卷积检测器(lightweight scale fair single convolution detector,LSFSCD)。相比传统检测方法和基于CNN的通用检测等方法,其架构更加简单,模型更小。该架构减少了误检和错检,实现更高检测精度的同时减少训练时间。通过使用Caffe框架在8g显存GTX1080上对VEDAI和DLR数据集进行实验,其结果验证了所提算法的有效性。
To solve the shortcomings of the general target detection framework in natural scenes,such as the lack of small vehicle target detection performance in aerial image,a real-time detector for small vehicle targets dedicated to aerial imagery was proposed,which was a lightweight scale fair single convolution detector(LSFSCD).Compared with traditional detection methods and general detection methods based on CNN,the architecture is simpler and the model is smaller.This architecture reduces false detections and false positives,enabling higher detection accuracy and using less training time.The effectiveness of the proposed algorithm is verified by experimenting on VEDAI and DLR data sets on 8 g memory GTX1080 using Caffe framework.
To solve the shortcomings of the general target detection framework in natural scenes,such as the lack of small vehicle target detection performance in aerial image,a real-time detector for small vehicle targets dedicated to aerial imagery was proposed,which was a lightweight scale fair single convolution detector(LSFSCD).Compared with traditional detection methods and general detection methods based on CNN,the architecture is simpler and the model is smaller.This architecture reduces false detections and false positives,enabling higher detection accuracy and using less training time.The effectiveness of the proposed algorithm is verified by experimenting on VEDAI and DLR data sets on 8 g memory GTX1080 using Caffe framework.
引文
[1]CHEN Yun,WU Fei,JING Xiaoyuan.Object tracking via multi-task structured sparse model[J].Computer Engineering and Design,2016,37(6):1663-1667(in Chinese).[陈芸,吴飞,荆晓远.基于多任务结构化稀疏模型的目标跟踪[J].计算机工程与设计,2016,37(6):1663-1667.]
[2]WU Zhifang,LIU Xin.Traffic flow estimation and traffic state detection based on the space-time map[J].Computer Engineering and Science,2016,38(9):1849-1857(in Chinese).[吴志芳,刘昕.基于时空图的交通流量统计和交通状态检测[J].计算机工程与科学,2016,38(9):1849-1857.]
[3]Deng Z,Sun H,Zhou S,et al.Toward fast and accurate vehicle detection in aerial images using coupled region-based convolutional neural networks[J].IEEE Journal of Selected Topics in Applied Earth Observations&Remote Sensing,2017,10(8):3652-3664.
[4]Liu K,Mattyus G.Fast multiclass vehicle detection on aerial images[J].IEEE Geoscience&Remote Sensing Letters,2015,12(9):1938-1942.
[5]Moranduzzo T,Melgani F.Automatic car counting method for unmanned aerial vehicle images[J].IEEE Transactions on Geoscience&Remote Sensing,2013,52(3):1635-1647.
[6]Moranduzzo T,Melgani F.Detecting cars in UAV images with a catalog-based approach[J].IEEE Transactions on Geoscience&Remote Sensing,2014,52(10):6356-6367.
[7]Diao W,Sun X,Zheng X,et al.Efficient saliency-based object detection in remote sensing images using deep belief networks[J].IEEE Geoscience&Remote Sensing Letters,2016,13(2):137-141.
[8]Chen Z,Wang C,Wen C,et al.Vehicle detection in highresolution aerial images via sparse representation and superpixels[J].IEEE Transactions on Geoscience&Remote Sensing,2015,54(1):103-116.
[9]Sommer LW,Schuchert T,Beyerer J.Fast deep vehicle detection in aerial images[C]//Applications of Computer Vision,2017:311-319.
[10]Girshick R.Fast R-CNN[EB/OL].[2018-05-28].https://arxiv.org/pdf/1504.08083.pdf.
[11]Ren S,He K,Girshick R,et al.Faster R-CNN:Towards real-time object detection with region proposal networks[J].IEEE Transactions on Pattern Analysis&Machine Intelligence,2017,39(6):1137-1149.
[12]Yang MY,Liao W,Li X,et al.Vehicle detection in aerial images[EB/OL].[2018-05-28].https://arxiv.org/pdf/1801.07339.pdf.
[13]Girshick R,Donahue J,Darrell T,et al.Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2014:580-587.
[14]Kim KH,Hong S,Roh B,et al.PVANET:Deep but lightweight neural networks for real-time object detection[EB/OL].[2018-05-21].https://arxiv.org/pdf/1608.08021.pdf.
[15]Dai J,Li Y,He K,et al.R-fcn:Object detection via region-based fully convolutional networks[C]//Advances in Neural Information Processing Systems,2016:379-387.
[16]He K,Gkioxari G,Dollar P,et al.Mask R-CNN[C]//IEEE International Conference on Computer Vision,2017:2980-2988.
[17]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,2016:779-788.
[18]Redmon J,Farhadi A.YOLO9000:Better,Faster,Stronger[C]//IEEE International Conference on Computer Vision,2017:6517-6525.
[19]Liu W,Anguelov D,Erhan D,et al.SSD:Single shot MultiBox detector[C]//European Conference on Computer Vision,2016:21-37.
[20]Huang J,Rathod V,Sun C,et al.Speed/accuracy TradeOffs for modern convolutional object detectors[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2017:7310-7311.
[21]Najibi M,Samangouei P,Chellappa R,et al.Ssh:Single stage headless face detector[C]//Proceedings of the IEEEConference on Computer Vision and Pattern Recognition,2017:4875-4884.
[22]Simonyan K,Zisserman A.Very deep convolutional networks for large-scale image recognition[EB/OL].[2018-05-21].https://arxiv.org/pdf/1409.1556.pdf.
[23]Luo W,Li Y,Urtasun R,et al.Understanding the effective receptive field in deep convolutional neural networks[C]//Advances in Neural Information Processing Systems,2016:4898-4906.
[24]YANG Guoliang,XU Nan,KANG Lele,et al.Identification of navel orange lesions leaves based on parametric exponential non-linear residual neural network[J].ACTA Agriculturae Zhejiangensis,2018,30(6):1073-1081(in Chinese).[杨国亮,许楠,康乐乐,等.基于参数指数非线性残差神经网络的脐橙病变叶片识别[J].浙江农业学报,2018,30(6):1073-1081.]
[25]Razakarivony S,Jurie F.Vehicle detection in aerial imagery:A small target detection benchmark[J].Journal of Visual Communication&Image Representation,2016,34(1):187-203.
[2]WU Zhifang,LIU Xin.Traffic flow estimation and traffic state detection based on the space-time map[J].Computer Engineering and Science,2016,38(9):1849-1857(in Chinese).[吴志芳,刘昕.基于时空图的交通流量统计和交通状态检测[J].计算机工程与科学,2016,38(9):1849-1857.]
[3]Deng Z,Sun H,Zhou S,et al.Toward fast and accurate vehicle detection in aerial images using coupled region-based convolutional neural networks[J].IEEE Journal of Selected Topics in Applied Earth Observations&Remote Sensing,2017,10(8):3652-3664.
[4]Liu K,Mattyus G.Fast multiclass vehicle detection on aerial images[J].IEEE Geoscience&Remote Sensing Letters,2015,12(9):1938-1942.
[5]Moranduzzo T,Melgani F.Automatic car counting method for unmanned aerial vehicle images[J].IEEE Transactions on Geoscience&Remote Sensing,2013,52(3):1635-1647.
[6]Moranduzzo T,Melgani F.Detecting cars in UAV images with a catalog-based approach[J].IEEE Transactions on Geoscience&Remote Sensing,2014,52(10):6356-6367.
[7]Diao W,Sun X,Zheng X,et al.Efficient saliency-based object detection in remote sensing images using deep belief networks[J].IEEE Geoscience&Remote Sensing Letters,2016,13(2):137-141.
[8]Chen Z,Wang C,Wen C,et al.Vehicle detection in highresolution aerial images via sparse representation and superpixels[J].IEEE Transactions on Geoscience&Remote Sensing,2015,54(1):103-116.
[9]Sommer LW,Schuchert T,Beyerer J.Fast deep vehicle detection in aerial images[C]//Applications of Computer Vision,2017:311-319.
[10]Girshick R.Fast R-CNN[EB/OL].[2018-05-28].https://arxiv.org/pdf/1504.08083.pdf.
[11]Ren S,He K,Girshick R,et al.Faster R-CNN:Towards real-time object detection with region proposal networks[J].IEEE Transactions on Pattern Analysis&Machine Intelligence,2017,39(6):1137-1149.
[12]Yang MY,Liao W,Li X,et al.Vehicle detection in aerial images[EB/OL].[2018-05-28].https://arxiv.org/pdf/1801.07339.pdf.
[13]Girshick R,Donahue J,Darrell T,et al.Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2014:580-587.
[14]Kim KH,Hong S,Roh B,et al.PVANET:Deep but lightweight neural networks for real-time object detection[EB/OL].[2018-05-21].https://arxiv.org/pdf/1608.08021.pdf.
[15]Dai J,Li Y,He K,et al.R-fcn:Object detection via region-based fully convolutional networks[C]//Advances in Neural Information Processing Systems,2016:379-387.
[16]He K,Gkioxari G,Dollar P,et al.Mask R-CNN[C]//IEEE International Conference on Computer Vision,2017:2980-2988.
[17]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,2016:779-788.
[18]Redmon J,Farhadi A.YOLO9000:Better,Faster,Stronger[C]//IEEE International Conference on Computer Vision,2017:6517-6525.
[19]Liu W,Anguelov D,Erhan D,et al.SSD:Single shot MultiBox detector[C]//European Conference on Computer Vision,2016:21-37.
[20]Huang J,Rathod V,Sun C,et al.Speed/accuracy TradeOffs for modern convolutional object detectors[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2017:7310-7311.
[21]Najibi M,Samangouei P,Chellappa R,et al.Ssh:Single stage headless face detector[C]//Proceedings of the IEEEConference on Computer Vision and Pattern Recognition,2017:4875-4884.
[22]Simonyan K,Zisserman A.Very deep convolutional networks for large-scale image recognition[EB/OL].[2018-05-21].https://arxiv.org/pdf/1409.1556.pdf.
[23]Luo W,Li Y,Urtasun R,et al.Understanding the effective receptive field in deep convolutional neural networks[C]//Advances in Neural Information Processing Systems,2016:4898-4906.
[24]YANG Guoliang,XU Nan,KANG Lele,et al.Identification of navel orange lesions leaves based on parametric exponential non-linear residual neural network[J].ACTA Agriculturae Zhejiangensis,2018,30(6):1073-1081(in Chinese).[杨国亮,许楠,康乐乐,等.基于参数指数非线性残差神经网络的脐橙病变叶片识别[J].浙江农业学报,2018,30(6):1073-1081.]
[25]Razakarivony S,Jurie F.Vehicle detection in aerial imagery:A small target detection benchmark[J].Journal of Visual Communication&Image Representation,2016,34(1):187-203.