[1]陈聪,杨忠,宋佳蓉,等.一种改进的卷积神经网络行人识别方法[J].应用科技,2019,46(03):51-57.[doi:10.11991/yykj.201809014]
 CHEN Cong,YANG Zhong,SONG Jiarong,et al.An improved pedestrian detection method based on convolutional neural network[J].Applied science and technology,2019,46(03):51-57.[doi:10.11991/yykj.201809014]
点击复制

一种改进的卷积神经网络行人识别方法(/HTML)
分享到:

《应用科技》[ISSN:1009-671X/CN:23-1191/U]

卷:
第46卷
期数:
2019年03期
页码:
51-57
栏目:
自动化技术
出版日期:
2019-04-29

文章信息/Info

Title:
An improved pedestrian detection method based on convolutional neural network
作者:
陈聪 杨忠 宋佳蓉 韩家明
南京航空航天大学 自动化学院, 江苏 南京 211106
Author(s):
CHEN Cong YANG Zhong SONG Jiarong HAN Jiaming
College of Automation Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China
关键词:
行人检测卷积神经网络深度学习YOLO特征提取聚类分析多尺度特征行人数据集
Keywords:
pedestrian detectionconvolutional neural network(CNN)deep learningYOLOfeature extractioncluster analysismulti-scale featurepedestrian dataset
分类号:
TP751
DOI:
10.11991/yykj.201809014
文献标志码:
A
摘要:
针对现有的行人检测算法存在的定位精度低、实时性差的问题,借鉴目标检测的研究成果You Only Look Once(YOLO)算法,提出一种实时的行人检测方法。以Tiny-YOLO为基础,改变网络模型的输入尺寸,获得更好的行人特征表达;结合图像中行人尺寸特点,使用聚类分析方法,对数据集进行目标框聚类,选取适合行人检测的候选框尺寸与数量;通过增加一定数量卷积层的方法重新设计特征提取和目标检测网络;在混合数据集上训练,增强模型泛化性。实验结果表明,在应对不同尺寸行人和部分遮挡情况时,文中方法具有更低的漏检率、更好的定位精度与检测效果,且检测速度可以满足实时性要求。
Abstract:
To solve the problem of poor positioning accuracy and low real-time performance in current pedestrian detection algorithms, a real-time pedestrian detection method is proposed by applying the You Only Look Once (YOLO) algorithm of target detection. Based on the Tiny-YOLO network, by changing the size of input of the network model, it helps the network gain better expression of pedestrian characteristics. Combined with the pedestrian size characteristics in the image, the cluster analysis method is used to calculate the initial candidate boxes in dataset, and the size and number of anchor boxes suitable for pedestrian detection are selected. Convolution layers are added in order to redesign the feature extraction and target detection networks. Train the network on a mixed dataset to improve the generalization of the model. The experimental results show that the method has lower miss rate, better positioning accuracy and detection effect when it comes to the situation of different sized pedestrians and partial occlusion, and the detection speed can meet the real-time requirements.

参考文献/References:

[1] SABZMEYDANI P, MORI G. Detecting pedestrians by learning shapelet features[C]//Proceedings of 2007 IEEE Conference on Computer Vision and Pattern Recognition. Minneapolis, MN, USA:IEEE, 2007:1-8.
[2] WANG Xiaoyu, HAN T X, YAN Shuicheng. An HOG-LBP human detector with partial occlusion handling[C]//Proceedings of the 12th International Conference on Computer Vision. Kyoto:IEEE Press, 2009:32-39.
[3] 高宗, 李少波, 陈济楠, 等. 基于YOLO网络的行人检测方法[J]. 计算机工程, 2018, 44(5):215-219
[4] GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of 2014 IEEE Conference on Computer Vision and Pattern Recognition. Columbus, OH, USA:IEEE, 2014.
[5] REN Shaoqing, HE Kaiming, GIRSHICK R, et al. Faster R-CNN:towards real-time object detection with region proposal networks[C]//Proceedings of the 28th International Conference on Neural Information Processing Systems. Montreal, Canada:ACM, 2015:91-99.
[6] LIU Wei, ANGUELOV D, ERHAN D, et al. SSD:single shot multibox detector[C]//Proceedings of the 14th European Conference on Computer Vision. Amsterdam:Springer, 2016:21-37.
[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. Las Vegas, NV, USA:IEEE, 2016:779-788.
[8] REDMON J, FARHADI A. YOLO9000:better, faster, stronger[C]//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, HI, USA:IEEE, 2017:6517-6525.
[9] REDMON J, FARHADI A. YOLOv3:an incremental improvement[J]. arXiv:1804.02767, 2018.
[10] REDMON J. Darknet:open source neural networks in C[DB/OL].[2017-10-25]. http://pjreddie.com/darknet/.
[11] 刘辉, 彭力, 闻继伟. 基于改进全卷积网络的多尺度感知行人检测算法[J]. 激光与光电子学进展, 2018, 55(9):91504
[12] 冯帅, 张龙, 贺小慧. 基于Jetson TK1和深度卷积神经网络的行人检测[J]. 信息技术, 2017(10):62-64
[13] MAO Jiayuan, XIAO Tete, JIANG Yuning, et al. What can help pedestrian detection?[C]//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, HI, USA:IEEE, 2017:6034-6043.
[14] 郝旭政, 柴争义. 一种改进的深度残差网络行人检测方法[J/OL]. 计算机应用研究, 2019, 36(6).[2018-05-22]. (2018-03-16) http://www.arocmag.com/article/02-2019-06-057.html. DOI:10.3969/j.issn.1001-3695.2017.12.0836.
[15] DAI Jifeng, LI Yi, HE Kaiming, et al. R-FCN:object detection via region-based fully convolutional networks[C]//Proceedings of the 30th International Conference on Neural Information Processing Systems. Barcelona, Spain:ACM, 2016:379-387.
[16] HE Kaiming, ZHANG Xiangyu, REN Shaoqing, et al. Deep residual learning for image recognition[C]//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, NV, USA:IEEE, 2016:770-778.
[17] HUANG Gao, LIU Zhuang, VAN DER MAATEN L, et al. Densely connected convolutional networks[C]//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, HI, USA:IEEE, 2017:2261-2269.
[18] 刘建国, 罗杰, 王帅帅, 等. 基于YOLOv2的行人检测方法研究[J]. 数字制造科学, 2018, 16(1):50-54
[19] DOLLAR P, WOJEK C, SCHIELE B, et al. Pedestrian detection:an evaluation of the state of the art[J]. IEEE transactions on pattern analysis and machine intelligence, 2012, 34(4):743-761.
[20] DALAL N, TRIGGS B. Histograms of oriented gradients for human detection[C]//Proceedings of 2015 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. San Diego, CA, USA:IEEE Computer Society, 2005:886-893.
[21] FELZENSZWALB P F, GIRSHICK R B, MCALLESTER D, et al. Object detection with discriminatively trained part-based models[J]. IEEE transactions on pattern analysis and machine intelligence, 2010, 32(9):1627-1645.

相似文献/References:

[1]王宝玉,等.融合人脸与步态周期模式的行人检测[J].应用科技,2012,39(01):44.[doi:10.3969/j.issn.1009-671X.201110010]
 WANG Baoyu,BEN Xianye,WANG Kejun.A pedestrian detection method integrating face information and gait period pattern[J].Applied science and technology,2012,39(03):44.[doi:10.3969/j.issn.1009-671X.201110010]
[2]宋佳蓉,杨忠,张天翼,等.基于卷积神经网络和多类SVM的交通标志识别[J].应用科技,2018,45(05):71.[doi:10.11991/yykj.201710009]
 SONG Jiarong,YANG Zhong,ZHANG Tianyi,et al.Traffic signs recogniton based on convolutional neural networks and multi-class SVM[J].Applied science and technology,2018,45(03):71.[doi:10.11991/yykj.201710009]
[3]郜雨桐,宁慧,王巍,等.基于卷积神经网络的车辆型号识别研究[J].应用科技,2018,45(06):53.[doi:10.11991/yykj.201803011]
 GAO Yutong,NING Hui,WANG Wei,et al.Research on vehicle model identification based on convolutional neural network[J].Applied science and technology,2018,45(03):53.[doi:10.11991/yykj.201803011]
[4]王立新,江加和.基于深度学习的显著性区域的图像检索研究[J].应用科技,2018,45(06):63.[doi:10.11991/yykj.201803012]
 WANG Lixin,JIANG Jiahe.Research on image retrieval of saliency region based on deep learning[J].Applied science and technology,2018,45(03):63.[doi:10.11991/yykj.201803012]
[5]卢用煌,黄山.深度学习在身份证号码识别中的应用[J].应用科技,2019,46(01):123.[doi:10.11991/yykj.201804008]
 LU Yonghuang,HUANG Shan.Application of deep learning in identification of ID card number[J].Applied science and technology,2019,46(03):123.[doi:10.11991/yykj.201804008]

备注/Memo

备注/Memo:
收稿日期:2018-09-14。
基金项目:国家自然科学基金项目(61473144);江苏高校优势学科建设工程资助项目;中国南方电网有限责任公司科技项目(066600KK52170074)
作者简介:陈聪,男,硕士研究生;杨忠,男,教授,博士生导师
通讯作者:杨忠,E-mail:YangZhong@nuaa.edu.cn
更新日期/Last Update: 2019-04-29