基于CapsNet的行人检测方法及评价
|
摘要
为了提高道路环境中行人目标检测的准确率,改善现有检测算法对不同环境视角下漏检率较高、耗时过长、实用性较差等问题,本文提出了一种基于CapsNet的行人检测模型.CapsNet由神经元所构成的Capsule组成,通过动态路由协议对物体的实例化参数进行表达和传递,保留了各特征对象间的空间层级,采用Caltech公开数据库对所提算法的有效性进行验证,并在检测准确率及算法耗时等方面与其他算法进行对比.实验结果表明:相比于其他主流检测算法,本文算法在确保检测效率的前提下,对数平均漏检率最低可降至9.17%;且在Caltech、INRIA和NICTA数据集的交叉验证实验中,也能达到良好的检测效果,具有较好的鲁棒性和泛化能力.
To enhance the detection accuracy of pedestrian target in the road environment, and solve the problems in the existing detection algorithms such as high undetected rate, long time-consumption and poor practicability under different environmental perspectives, a CapsNet-based pedestrian detection model is presented in this paper.CapsNet, consisted of a Capsule composed of neurons, expresses and transfers the instantiation parameters of the object through the dynamic routing protocol while preserving the spatial hierarchy between the feature objects,verifies the validity of the proposed algorithm based on the Caltech public database, and compares with other algorithms in terms of detection accuracy and time consumption of the algorithms. The experimental results show that compared with other mainstream detection algorithms, the algorithm can reduce the undetected rate to a minimum of 9.17% under the premise of ensuring the detection efficiency, and in the cross-validation experiment of Caltech, INRIA and NICTA data sets, good detection results can be achieved, with remarkable robustness and generalization ability.
To enhance the detection accuracy of pedestrian target in the road environment, and solve the problems in the existing detection algorithms such as high undetected rate, long time-consumption and poor practicability under different environmental perspectives, a CapsNet-based pedestrian detection model is presented in this paper.CapsNet, consisted of a Capsule composed of neurons, expresses and transfers the instantiation parameters of the object through the dynamic routing protocol while preserving the spatial hierarchy between the feature objects,verifies the validity of the proposed algorithm based on the Caltech public database, and compares with other algorithms in terms of detection accuracy and time consumption of the algorithms. The experimental results show that compared with other mainstream detection algorithms, the algorithm can reduce the undetected rate to a minimum of 9.17% under the premise of ensuring the detection efficiency, and in the cross-validation experiment of Caltech, INRIA and NICTA data sets, good detection results can be achieved, with remarkable robustness and generalization ability.
引文
[1] HOSANG J, BENENSON R, DOLLAR P, et al. What makes for effective detection proposals[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(4):814-830.
[2]种衍文,匡湖林,李清泉.一种基于多特征和机器学习的分级行人检测方法[J].自动化学报, 2012, 38(3):375-381.[ZHONG Y W, KUANG H L, LI Q Q. Twostage pedestrian detection based on multiple features and machine learning[J]. Acta. Automatica Sinica, 2012,38(3):375-381.]
[3] LI X F, LI L X, FLOHR F, et al. A unified framework for concurrent pedestrian and cyclist detection[J]. IEEE Transactions on Intelligent Transportation Systems,2017, 18(2):269-281.
[4]邵春福,李娟,赵熠,等.行人交通的视频检测方法综述[J].交通运输系统工程与信息, 2008, 8(4):23-29.[SHAO C F, LI J, ZHAO Y, et al. Review of pedestrian traffic data collection method based on video image processing[J]. Journal of Transportation Systems Engineering and Information Technology, 2008, 8(4):23-29.]
[5] GIRSHICK R, DONAHUE J, DARRELL T, et al.Region-based convolutional networks for accurate object detection and segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(1):142-158.
[6] DOLLAR P, WOJEK C, SCHIELE B, et al. Pedestrian detection:An evaluation of state of the art[J]. IEEE Transactions on Pattern Analysis&Machine Intelligence, 2012, 34(4):743-761.
[7]王峰萍,王卫星,杨楠,等.基于改进Retinex的城市交通图像增强[J].交通运输系统工程与信息, 2017, 17(5):53-59.[WANG F P, WANG W X, YANG N, et al.An urban traffic image enhancement method based on modified retinex[J]. Journal of Transportation Systems Engineering and Information Technology, 2017, 17(5):53-59.]
[8]王春瑶,陈俊周,李炜.超像素分割算法研究综述[J].计算机应用研究, 2014, 31(1):6-12.[WANG C Y,CHEN J Z, LI W. Review on superpixel segmentation algorithms[J]. Application Research of Computers,2014, 31(1):6-12.]
[9] VIOLA P, JONES M J, SNOW D. Detecting pedestrians using patterns of motion and appearance[J].International Journal of Computer Vision, 2005, 63(2):153-161.
[10] VIOLA P, JONES M J. Robust real-time face detection[J]. International Journal of Computer Vision, 2004, 57(2):137-154.
[11] DOLLAR P, APPEL R, BELONGIE S, et al. Fast feature pyramids for object detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014, 36(8):1532-1545.
[12] OUYANG W L, WANG X. Joint deep learning for pedestrian detection[C]//IEEE International Conference on Computer Vision, Sydney:IEEE, 2014:2056-2063.
[13] HOSANG J, OMRAN M, BENENSON R, et al. Taking a deeper look at pedestrians[C]//Computer Vision and Pattern Recognition, Boston:IEEE, 2015:4073-4082.
[14] CAI Z, SABERIAN M, VASCONCELOS N. Learning complexity-aware cascades for deep pedestrian detection[C]//IEEE International Conference on Computer Vision, Santiago:IEEE, 2015:3361-3369.
[15] LI J, LIANG X, SHEN S M, et al. Scale-aware Fast RCNN for pedestrian detection[J]. IEEE Transactions on Multimedia, 2015, 20(4):985-996.
[16] MAJI S, BERG A C, MALIK J. Efficient classification for additive kernel SVMs[J]. IEEE Transactions on Pattern Analysis&Machine Intelligence, 2012, 35(1):66-77.
[2]种衍文,匡湖林,李清泉.一种基于多特征和机器学习的分级行人检测方法[J].自动化学报, 2012, 38(3):375-381.[ZHONG Y W, KUANG H L, LI Q Q. Twostage pedestrian detection based on multiple features and machine learning[J]. Acta. Automatica Sinica, 2012,38(3):375-381.]
[3] LI X F, LI L X, FLOHR F, et al. A unified framework for concurrent pedestrian and cyclist detection[J]. IEEE Transactions on Intelligent Transportation Systems,2017, 18(2):269-281.
[4]邵春福,李娟,赵熠,等.行人交通的视频检测方法综述[J].交通运输系统工程与信息, 2008, 8(4):23-29.[SHAO C F, LI J, ZHAO Y, et al. Review of pedestrian traffic data collection method based on video image processing[J]. Journal of Transportation Systems Engineering and Information Technology, 2008, 8(4):23-29.]
[5] GIRSHICK R, DONAHUE J, DARRELL T, et al.Region-based convolutional networks for accurate object detection and segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(1):142-158.
[6] DOLLAR P, WOJEK C, SCHIELE B, et al. Pedestrian detection:An evaluation of state of the art[J]. IEEE Transactions on Pattern Analysis&Machine Intelligence, 2012, 34(4):743-761.
[7]王峰萍,王卫星,杨楠,等.基于改进Retinex的城市交通图像增强[J].交通运输系统工程与信息, 2017, 17(5):53-59.[WANG F P, WANG W X, YANG N, et al.An urban traffic image enhancement method based on modified retinex[J]. Journal of Transportation Systems Engineering and Information Technology, 2017, 17(5):53-59.]
[8]王春瑶,陈俊周,李炜.超像素分割算法研究综述[J].计算机应用研究, 2014, 31(1):6-12.[WANG C Y,CHEN J Z, LI W. Review on superpixel segmentation algorithms[J]. Application Research of Computers,2014, 31(1):6-12.]
[9] VIOLA P, JONES M J, SNOW D. Detecting pedestrians using patterns of motion and appearance[J].International Journal of Computer Vision, 2005, 63(2):153-161.
[10] VIOLA P, JONES M J. Robust real-time face detection[J]. International Journal of Computer Vision, 2004, 57(2):137-154.
[11] DOLLAR P, APPEL R, BELONGIE S, et al. Fast feature pyramids for object detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014, 36(8):1532-1545.
[12] OUYANG W L, WANG X. Joint deep learning for pedestrian detection[C]//IEEE International Conference on Computer Vision, Sydney:IEEE, 2014:2056-2063.
[13] HOSANG J, OMRAN M, BENENSON R, et al. Taking a deeper look at pedestrians[C]//Computer Vision and Pattern Recognition, Boston:IEEE, 2015:4073-4082.
[14] CAI Z, SABERIAN M, VASCONCELOS N. Learning complexity-aware cascades for deep pedestrian detection[C]//IEEE International Conference on Computer Vision, Santiago:IEEE, 2015:3361-3369.
[15] LI J, LIANG X, SHEN S M, et al. Scale-aware Fast RCNN for pedestrian detection[J]. IEEE Transactions on Multimedia, 2015, 20(4):985-996.
[16] MAJI S, BERG A C, MALIK J. Efficient classification for additive kernel SVMs[J]. IEEE Transactions on Pattern Analysis&Machine Intelligence, 2012, 35(1):66-77.
© 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 |