高级搜索

基于特征通道和空间联合注意机制的遮挡行人检测方法

陈勇 刘曦 刘焕淋

引用本文: 陈勇, 刘曦, 刘焕淋. 基于特征通道和空间联合注意机制的遮挡行人检测方法[J]. 电子与信息学报, doi: 10.11999/JEIT190606 shu
Citation:  Yong CHEN, Xi LIU, Huanlin LIU. Occluded Pedestrian Detection Based on Joint Attention Mechanism of Channel-wise and Spatial Information[J]. Journal of Electronics and Information Technology, doi: 10.11999/JEIT190606 shu

基于特征通道和空间联合注意机制的遮挡行人检测方法

    作者简介: 陈勇: 男,1963年生,博士,教授,研究方向为图像处理;
    刘曦: 男,1993年生,硕士生,研究方向为行人目标检测;
    刘焕淋: 女,1970年生,博士,教授,研究方向为信号处理等方面的研究
    通讯作者: 陈勇,chenyong@cqupt.edu.cn
  • 基金项目: 国家自然科学基金(51977021)

摘要: 遮挡是行人检测任务中导致漏检发生的主要原因之一,对检测器性能造成了不利影响。为了增强检测器对于遮挡行人目标的检测能力,该文提出一种基于特征引导注意机制的单级行人检测方法。首先,设计一种特征引导注意模块,在保持特征通道间的关联性的同时保留了特征图的空间信息,引导模型关注遮挡目标可视区域;然后,通过注意模块融合浅层和深层特征,从而提取到行人的高层语义特征;最后,将行人检测作为一种高层语义特征检测问题,通过激活图的形式预测得到行人位置和尺度,并生成最终的预测边界框,避免了基于先验框的预测方式所带来的额外参数设置。所提方法在CityPersons数据集上进行了测试,并在Caltech数据集上进行了跨数据集实验。结果表明该方法对于遮挡目标检测准确度优于其他对比算法。同时该方法实现了较快的检测速度,取得了检测准确度和速度的平衡。

English

    1. [1]

      张功国, 吴建, 易亿, 等. 基于集成卷积神经网络的交通标志识别[J]. 重庆邮电大学学报: 自然科学版, 2019, 31(4): 571–577. doi: 10.3979/j.issn.1673-825X.2019.04.019
      ZHANG Gongguo, WU Jian, YI Yi, et al. Traffic sign recognition based on ensemble convolutional neural network[J]. Journal of Chongqing University of Posts and Telecommunications:Natural Science Edition, 2019, 31(4): 571–577. doi: 10.3979/j.issn.1673-825X.2019.04.019

    2. [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. doi: 10.3724/SP.J.1004.2012.00375
      CHONG Yanwen, KUANG Hulin, and LI Qingquan. Two-stage pedestrian detection based on multiple features and machine learning[J]. Acta Automatica Sinica, 2012, 38(3): 375–381. doi: 10.3724/SP.J.1004.2012.00375

    3. [3]

      刘威, 段成伟, 遇冰, 等. 基于后验HOG特征的多姿态行人检测[J]. 电子学报, 2015, 43(2): 217–224. doi: 10.3969/j.issn.0372-2112.2015.02.002
      LIU Wei, DUAN Chengwei, YU Bing, et al. Multi-pose pedestrian detection based on posterior HOG feature[J]. Acta Electronica Sinica, 2015, 43(2): 217–224. doi: 10.3969/j.issn.0372-2112.2015.02.002

    4. [4]

      REN Shaoqing, HE Kaiming, GIRSHICK R, et al. Faster R-CNN: Towards real-time object detection with region proposal networks[C]. 2015 Advances in Neural Information Processing Systems, Montreal, Canada, 2015: 91–99.

    5. [5]

      LI Jianan, LIANG Xiaodan, SHEN Shengmei, et al. Scale-aware fast R-CNN for pedestrian detection[J]. IEEE Transactions on Multimedia, 2018, 20(4): 985–996. doi: 10.1109/TMM.2017.2759508

    6. [6]

      王进, 陈知良, 李航, 等. 一种基于增量式超网络的多标签分类方法[J]. 重庆邮电大学学报: 自然科学版, 2019, 31(4): 538–549. doi: 10.3979/j.issn.1673-825X.2019.04.015
      WANG Jin, CHEN Zhiliang, LI Hang, et al. Hierarchical multi-label classification using incremental hypernetwork[J]. Journal of Chongqing University of Posts and Telecommunications:Natural Science Edition, 2019, 31(4): 538–549. doi: 10.3979/j.issn.1673-825X.2019.04.015

    7. [7]

      GIRSHICK R. Fast R-CNN[C]. 2015 IEEE International Conference on Computer Vision, Santiago, Chile, 2015: 1440–1448.

    8. [8]

      LIU Wei, LIAO Shengcai, HU Weidong, et al. Learning efficient single-stage pedestrian detectors by asymptotic localization fitting[C]. The 15th European Conference on Computer Vision, Munich, Germany, 2018: 618–634.

    9. [9]

      LIU Wei, LIAO Shengcai, REN Weiqiang, et al. High-level semantic feature detection: A new perspective for pedestrian detection[C]. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, USA, 2019: 5182–5191.

    10. [10]

      ZHANG Shanshan, BENENSON R, OMRAN M, et al. How far are we from solving pedestrian detection?[C]. 2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, USA, 2016: 1259–1267.

    11. [11]

      ZHANG Shifeng, WEN Longyin, BIAN Xiao, et al. Occlusion-aware R-CNN: detecting pedestrians in a crowd[C]. The 15th European Conference on Computer Vision, Munich, Germany, 2018: 637–653.

    12. [12]

      OUYANG Wanli, ZHOU Hui, LI Hongsheng, et al. Jointly learning deep features, deformable parts, occlusion and classification for pedestrian detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, 40(8): 1874–1887. doi: 10.1109/TPAMI.2017.2738645

    13. [13]

      FEI Chi, LIU Bin, CHEN Zhu, et al. Learning pixel-level and instance-level context-aware features for pedestrian detection in crowds[J]. IEEE Access, 2019, 7: 94944–94953. doi: 10.1109/ACCESS.2019.2928879

    14. [14]

      LIN C Y, XIE Hongxia, and ZHENG Hua. PedJointNet: Joint head-shoulder and full body deep network for pedestrian detection[J]. IEEE Access, 2019, 7: 47687–47697. doi: 10.1109/ACCESS.2019.2910201

    15. [15]

      ZHANG Shanshan, YANG Jian, and SCHIELE B. Occluded pedestrian detection through guided attention in CNNs[C]. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, USA, 2018: 6995–7003.

    16. [16]

      ZHU Chenchen, HE Yihui, and SAVVIDES M. Feature selective anchor-free module for single-shot object detection[C]. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, USA, 2019: 840–849.

    17. [17]

      LIN T Y, DOLLÁR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, USA, 2017: 936–944.

    18. [18]

      CHEN Long, ZHANG Hanwang, XIAO Jun, et al. SCA-CNN: Spatial and channel-wise attention in convolutional networks for image captioning[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, USA, 2017: 6298–6306.

    19. [19]

      WOO S, PARK J, LEE J Y, et al. Cbam: Convolutional block attention module[C]. The 15th European Conference on Computer Vision, Munich, Germany, 2018: 3–19.

    20. [20]

      SONG Tao, SUN Leiyu, XIE Di, et al. Small-scale pedestrian detection based on topological line localization and temporal feature aggregation[C]. The 15th European Conference on Computer Vision, Munich, Germany, 2018: 536–551.

    21. [21]

      ZHANG Shanshan, BENENSON R, and SCHIELE B. Citypersons: A diverse dataset for pedestrian detection[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, USA, 2017: 4457–4465.

    22. [22]

      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. doi: 10.1109/TPAMI.2011.155

    23. [23]

      WANG Xinlong, XIAO Tete, JIANG Yuning, et al. Repulsion loss: Detecting pedestrians in a crowd[C]. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, USA, 2018: 7774–7783.

    1. [1]

      蔡靖, 刘峰, 丁润涛. 模糊迭代边缘检测器. 电子与信息学报,

    2. [2]

      胡军锋, 张海林, 杨远. 稳健的高效MIMO检测器. 电子与信息学报,

    3. [3]

      欧阳毅, 张三元, 张引. 基于窗口边缘梯度势能的人体遮挡多尺度检测算法. 电子与信息学报,

    4. [4]

      李向吉, 丁润涛, 蔡靖. 边缘检测器性能的客观定量评价. 电子与信息学报,

    5. [5]

      王鞠庭, 江胜利, 何劲, 刘中. 机载MIMO雷达广义最大似然检测器. 电子与信息学报,

    6. [6]

      阎石, 吕振肃. 基于进化算法的多用户检测器. 电子与信息学报,

    7. [7]

      孟祥伟. 非参数秩和检测器的性能分析. 电子与信息学报,

    8. [8]

      陆林根. 相关高斯噪声自适应非参量检测器. 电子与信息学报,

    9. [9]

      孙锐, 陈军, 高隽. 基于显著性检测与HOG-NMF特征的快速行人检测方法. 电子与信息学报,

    10. [10]

      张菁, 沈兰荪, 高静静. 基于视觉注意模型和进化规划的感兴趣区检测方法. 电子与信息学报,

    11. [11]

      刘龙, 孙强, 宋琦军. 面向目标检测的多尺度运动注意力融合算法研究. 电子与信息学报,

    12. [12]

      张力, 钱恭斌, 肖薇薇, 纪震. 基于双通道检测的多比特盲水印最优检测器的设计. 电子与信息学报,

    13. [13]

      邹鲲, 廖桂生, 李军, 李伟, 李天星. 非高斯杂波下知识辅助检测器敏感性分析. 电子与信息学报,

    14. [14]

      孟祥伟. 秩和非参数检测器在杂波边缘中的性能. 电子与信息学报,

    15. [15]

      陈其铭, 尹俊勋. 垂直分层空时码的MAP检测器及其神经网络实现. 电子与信息学报,

    16. [16]

      杨海峰, 蒋国喜, 刘维建, 谢文冲, 王永良. 零训练样本的集中式MIMO雷达斜对称检测器. 电子与信息学报,

    17. [17]

      杨海峰, 谢文冲, 唐瑭, 罗玉文, 刘维建, 王永良. 目标引入干扰条件下的Wald检测器. 电子与信息学报,

    18. [18]

      谢洪森, 邹鲲. 一种非均匀场景复合高斯杂波下的自适应检测器. 电子与信息学报,

    19. [19]

      何友, 关键, HermannRohling. 一种新的最大选择恒虚警检测器. 电子与信息学报,

    20. [20]

      顾新锋, 简涛, 何友. 距离扩展目标的双门限恒虚警检测器及性能分析. 电子与信息学报,

  • 图 1  模型总体结构

    图 2  注意模块总体结构

    图 3  特征通道注意模块结构

    图 4  空间关注模块

    图 5  行人解析网络

    图 6  可视化位置预测热图

    图 7  Caltech跨数据库实验

    表 1  验证实验条件设置

    R (Reasonable)HO (Heavy occlusion)R+HO (Reasonable+Heavy occlusion)
    $v \in [0.65,\infty )$$v \in [0.20,0.65]$$v \in [0.20,\infty )$
    下载: 导出CSV

    表 2  注意网络验证结果MR–2(%)

    方法RHOR+HO
    文献[16]16.056.738.2
    Baseline12.141.138.1
    Baseline+CA11.839.237.8
    Baseline+CA+SA11.638.537.3
    下载: 导出CSV

    表 3  CityPersons数据集测试结果MR-2(%)

    方法主干网络ReasonableHeavyPartialBare测试时间(s)
    OR-CNN[11]VGG-1612.855.715.36.7
    FasterRCNN[21]VGG-1615.4
    ALFNet[8]ResNet-5012.051.911.48.40.27
    CSP[9]ResNet-5011.049.310.47.30.33
    CAFL[13]ResNet-5011.450.412.17.6
    PedJointNet[14]ResNet-5013.552.1
    TLL[20]ResNet-5015.553.617.210.0
    RepLoss[23]ResNet-5013.256.916.87.6
    本文方法ResNet-5011.647.69.87.50.22
    下载: 导出CSV
  • 加载中
图(7)表(3)
计量
  • PDF下载量:  37
  • 文章访问数:  1392
  • HTML全文浏览量:  398
文章相关
  • 通讯作者:  陈勇, chenyong@cqupt.edu.cn
  • 收稿日期:  2019-08-09
  • 录用日期:  2020-02-18
  • 网络出版日期:  2020-03-13
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

/

返回文章