一种优化的Faster R-CNN小目标检测方法

doi:10.13474/j.cnki.11-2246.2021.0267

测绘通报 ›› 2021, Vol. 0 ›› Issue (9): 21-27.doi: 10.13474/j.cnki.11-2246.2021.0267

一种优化的Faster R-CNN小目标检测方法

程瑞¹, 高建¹, 邢强², 孙中昶²

1. 南京邮电大学, 江苏南京 210003;
2. 中国科学院空天信息创新研究院, 北京 100094

收稿日期:2020-09-02 修回日期:2021-03-19 出版日期:2021-09-25 发布日期:2021-10-11
通讯作者: 邢强。E-mail:xingqiang@aircas.ac.cn
作者简介:程瑞(1994-),男,硕士生,主要从事遥感数字图像处理。E-mail:1218012332@njupt.edu.cn
基金资助:
中国科学院战略性先导科技专项（XDA19090121；XDA19030104）

An optimized Faster R-CNN small target detection method

CHENG Rui¹, GAO Jian¹, XING Qiang², SUN Zhongchang²

1. College of telecommunications and Information Engineering, Nanjing University of Posts and Telecommunications, Nanjing 210003, China;
2. Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100094, China

Received:2020-09-02 Revised:2021-03-19 Online:2021-09-25 Published:2021-10-11

摘要/Abstract

摘要： 图像目标检测是计算机视觉与数字图像处理的一个热门方向，其主要任务是找出图像中感兴趣的物体并确定物体的位置与类别。目前基于深度学习模型是主流的目标检测算法，利用其解决诸多学科问题成为一种趋势。本文采用区域卷积神经网络（Faster R-CNN）深度学习算法和相关图像处理算法，以ResNet50、ResNet101为骨干网络，采用特征金字塔网络开展新冠疫情期间武汉市车辆变化监测，以此分析疫情下的武汉市内部活动强度。结果显示：本文车辆目标检测方法的精确率为0.96，召回率为0.92，平均精度为0.85。疫情前（2019年11月17日）、中（2020年02月22日）车辆变化情况为：武汉汇聚中心分别为263、32辆，汪家嘴立交桥分别为89、44辆，新兴工业园分别为554、347辆，经开未来城分别为188、57辆。可知，疫情导致武汉市人口出行减少、车辆活动明显降低。

关键词: 目标检测, Faster R-CNN算法, 深度学习, 图像处理, 新冠病毒

Abstract: Image object detection is a popular direction in computer vision and digital image processing. Its main task is to find out the object of interest in the image and determine the location and category of the object. The current mainstream object detection algorithms are mainly based on deep learning models, and it has become a trend to solve many disciplinary problems. This article uses a method based on the combination of the regional convolutional neural network (Faster R-CNN) deep learning algorithm and related image processing algorithms,using ResNet50 and ResNet101 as the backbone network and using feature pyramid networks to monitor the changes of vehicles in Wuhan during the new crown epidemic to analyze the intensity of internal activities in Wuhan during the epidemic. The results show that the accuracy rate of the image detection method in this paper is 0.96, the recall rate is 0.915, and the average accuracy is 0.853 8.The vehicle number changes before the epidemic (November 17, 2019) and during the epidemic (February 22, 2020) is as follows:Wuhan Convergence Center (263 and 32 vehicles), Wangjiazui Overpass (89 and 44 vehicles), Xinxing Industrial Park (554 and 347 vehicles), Jingkai Future City (188 and 57 vehicles). The epidemic has led to a decrease in population travel and vehicle activities in Wuhan.

Key words: target detection, Faster R-CNN algorithm, deep learning, image processing, novel coronavirus

中图分类号:

P237

程瑞, 高建, 邢强, 孙中昶. 一种优化的Faster R-CNN小目标检测方法[J]. 测绘通报, 2021, 0(9): 21-27.

CHENG Rui, GAO Jian, XING Qiang, SUN Zhongchang. An optimized Faster R-CNN small target detection method[J]. Bulletin of Surveying and Mapping, 2021, 0(9): 21-27.

参考文献

[1] KIM D J, LEE S H,SOHN M K. Face recognition via local directional pattern[J].International Journal of Security & Its Applications,2013,7(2):191-200.
[2] GONZALO D L T, PAUL R, KIM-KWANG R C. Driverless vehicle security:challenges and future research opportunities[J]. Future Generation Computer Systems,2018,10(8):1092-1111.
[3] LECUN Y, BOTTOU L, BENGIO Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998,86(11):2278-2324.
[4] KRIZHEVSKY A, SUTSKEVER I, HINTON G. ImageNet classification with deep convolutional neural networks[J]. Advances in Neural Information Processing Systems, 2012, 25(2):1097-1105.
[5] ZEILER M D, FERGUS R. Visualizing and understanding convolutional networks[C]//Proceedings of the 13th European Conference on Computer Vision. Berlin:Springer-Verlag, 2014:818-833.
[6] SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[C]//Proceedings of 2015 International Conference on Learning Representations. San Diego:ICLR,2015:1-5.
[7] REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once:unified, real-time object detection[C]//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition. Washington:IEEE Computer Society Press, 2016:429-442.
[8] LIU W, ANGUELOV D, ERHAN D, et al. SSD:single shot multibox detector[C]//Proceedings of 2016 Computer Vision-ECCV.[S.l.]:Springer, 2016:21-37.
[9] LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017(99):2999-3007.
[10] LI S X, ZHANG Z L, LI B. A plane target detection algorithm in remote sensing images based on deep learning network technology[J].Journal of Physics Conference, 2018, 9(6):20-25.
[11] 王凯,潘炼.基于改进Faster R-CNN图像小目标检测[J].电视技术,2019,43(20):77-80.
[12] 张子颖,王敏.基于Faster R-CNN和增量学习的车辆目标检测[J].计算机系统应用,2020,29(2):181-186.
[13] DINH V S, NGUYEN D M. Fully residual convolutional neural networks for aerial image segmentation[J]. So ICT,2018,289-296.
[14] KOSHY R, MAHMOOD A. Optimizing deep CNN architec-tures for face liveness detection[J]. Entropy, 2019,21(4):423-436.
[15] WU Z, AN H, JIN X, et al. Research and optimization of fast convolution algorithm winograd on intel platform[J].Journal of Computer Research and Development,2019,56(4):825-835.
[16] GUO W, LI W H, GONG W G, et al. Extended feature pyramid network with adaptive scale training strategy and anchors for object detection in aerial images[J]. Remote Sensing,2020,12(5):34-38.
[17] 王庆德,吴欣哲.大数据助力疫情防控和复工复产——基于新冠肺炎疫情[J].中国物价,2020(5):9-11.
[18] BARNETT E D, WALKER P F. Role of immigrants and migrants in emerging infectious diseases[J]. The Medical clinics of North America,2008,92(6):1447-1458.
[19] BYKAITLYN F, JOSH M. Satellite data suggests coron-avirus may have hit China earlier:Researchers[N].abcNews,2020-06-08.

一种优化的Faster R-CNN小目标检测方法

An optimized Faster R-CNN small target detection method

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	柳林, 孙毅, 李万武. 基于CNN海上钻井平台检测模型的构建及训练算法分析[J]. 测绘通报, 2022, 0(7): 26-32,99.
[2]	刘立, 董先敏, 刘娟, 文学虎. 人机融合智能的遥感解译生产新方法[J]. 测绘通报, 2022, 0(7): 118-123,137.
[3]	蔡柔丹. 一种基于用户异步轨迹的身份识别智能方法[J]. 测绘通报, 2022, 0(7): 158-162,167.
[4]	滕文鑫, 张建辰, 邵杰. 集成时空邻近与卷积网络车道级高精度定位算法[J]. 测绘通报, 2022, 0(6): 1-5,61.
[5]	文奴, 郭仁忠, 贺彪, 万远. YOLO v4框架下Multi-Patch多帧增量式交通视频目标检测[J]. 测绘通报, 2022, 0(5): 38-44.
[6]	张晋赫, 秦育罗, 张在岩, 宋伟东, 朱洪波. 复杂场景下农村道路裂缝分割方法[J]. 测绘通报, 2022, 0(5): 74-78,88.
[7]	师芸, 石龙龙, 牛敏杰, 赵侃. 基于单阶段实例分割网络的黄土滑坡多任务自动识别[J]. 测绘通报, 2022, 0(4): 26-31.
[8]	王雪, 梁珂, 隋立春, 钟棉卿, 朱剑锋. 膨胀卷积与金字塔表达的深度学习模型用于农村建筑物提取[J]. 测绘通报, 2022, 0(4): 61-65.
[9]	高鸣, 周鑫鑫, 刘琦, 杨光迪, 吴长彬. 深度学习网络支持下的农房侵占耕地自动化监测[J]. 测绘通报, 2022, 0(3): 47-53.
[10]	段晨阳, 冯建中, 全斌, 白林燕, 王盼盼. 利用深度学习进行GF-6影像枣园检测识别[J]. 测绘通报, 2022, 0(3): 54-59.
[11]	孙琴琴, 蔡国印, 杨柳忠, 张宁, 杜明义. 城市尺度典型地表要素综合提取方法研究[J]. 测绘通报, 2022, 0(2): 62-66.
[12]	贾永红. 数字图像处理混合教学的研究与实践[J]. 测绘通报, 2022, 0(2): 174-176.
[13]	张自强, 刘涛, 杜萍, 锁旭宏, 杨国林. 图残差神经网络支持下的建筑物群组模式分类[J]. 测绘通报, 2022, 0(1): 1-7.
[14]	车子杰, 高飞, 吴兆福, 李振轩. 基于改进U-Net网络的多源遥感影像洪涝灾害信息提取与变化分析[J]. 测绘通报, 2022, 0(1): 26-32.
[15]	淳锦, 张新长, 郭海京, 张建国, 金诗程. 基于多源信息与深度学习特征提取的人口空间抽样方法[J]. 测绘通报, 2021, 0(8): 42-47.