基于语义分割注意力与可见区域预测的行人检测方法

doi:10.12068/j.issn.1005-3026.2021.09.007

东北大学学报（自然科学版） ›› 2021, Vol. 42 ›› Issue (9): 1261-1267.DOI: 10.12068/j.issn.1005-3026.2021.09.007

基于语义分割注意力与可见区域预测的行人检测方法

王璐¹，王帅¹，张国峰¹，徐礼胜^{2, 3}

(1. 东北大学计算机科学与工程学院，辽宁沈阳110169; 2. 东北大学医学与生物信息工程学院，辽宁沈阳110169; 3. 沈阳东软智能医疗科技研究院有限公司，辽宁沈阳110167)

修回日期:2021-01-04 接受日期:2021-01-04 发布日期:2021-09-16
通讯作者: 王璐
作者简介:王璐(1980-)，女，辽宁沈阳人，东北大学副教授；徐礼胜(1975-)，男，安徽安庆人，东北大学教授，博士生导师.
基金资助:
中央高校基本科研业务费专项资金资助项目(N181604006); 辽宁省自然科学基金资助项目(20170540312); 国家自然科学基金资助项目(61773110); 沈阳市科学技术计划基金资助项目(20-201-4-10).

Pedestrian Detection Based on Semantic Segmentation Attention and Visible Region Prediction

WANG Lu¹， WANG Shuai¹， ZHANG Guo-feng¹， XU Li-sheng^2,3

1. School of Computer Science & Engineering， Northeastern University， Shenyang 110169， China; 2. School of Medicine and Biological Information Engineering， Northeastern University， Shenyang 110169， China; 3. Neusoft Research of Intelligent Healthcare Technology， Co.， Ltd.， Shenyang 110167， China.

Revised:2021-01-04 Accepted:2021-01-04 Published:2021-09-16
Contact: XU Li-sheng
About author:-
Supported by:
-

摘要/Abstract

摘要： 为改善图像中遮挡和小尺寸行人的检测精度，提出一种基于语义分割注意力和可见区域预测的行人检测方法.具体地，在SSD(single shot multi-box detector)目标检测网络的基础上，首先优化SSD的超参数设置，使其更适于行人检测;然后在主干网络中引入基于语义分割的注意力分支来增强行人检测特征的表达能力;最后提出一种检测预测模块，它不仅能同时预测行人整体和可见区域，还能利用可见区域预测分支所学的特征去引导整体检测特征的学习，提升检测效果.在Caltech行人检测数据集上进行了实验，所提方法的对数平均缺失率为5.5%，与已有方法相比具有一定的优势.

关键词: 行人检测;卷积神经网络;语义分割注意力;行人可见区域预测;多任务网络

Abstract: To improve the detection performance on occluded and small pedestrians in images， a pedestrian detection method based on semantic segmentation attention and visible region prediction was proposed. Specifically， based on the single shot multi-box detector(SSD)object detection network， the hyperparameter setting of the SSD was firstly optimized to make it more suitable for pedestrian detection. Then the semantic segmentation attention branch was introduced into the network to enhance the pedestrian detection features learned by the network. Finally， a detection prediction module which can simultaneously detect the full bodies and visible regions of pedestrians was developed. This module has the advantage of leveraging the features learned from visible regions to guide the learning of the full-body detection features， hence improving the overall detection accuracy. The experiment carried out on the Caltech pedestrian detection benchmark shows that the log-average miss rate of the proposed method is 5.5%， which is competitive compared with existing pedestrian detection approaches.

Key words: pedestrian detection; convolutional neural network; semantic segmentation attention(SSA); pedestrian visible region detection; multi-task network

中图分类号:

TP391.41

王璐，王帅，张国峰，徐礼胜. 基于语义分割注意力与可见区域预测的行人检测方法[J]. 东北大学学报（自然科学版）, 2021, 42(9): 1261-1267.

WANG Lu， WANG Shuai， ZHANG Guo-feng， XU Li-sheng. Pedestrian Detection Based on Semantic Segmentation Attention and Visible Region Prediction[J]. Journal of Northeastern University(Natural Science), 2021, 42(9): 1261-1267.

参考文献

[1]Wang L，Yung N.Extraction of moving objects from their background based on multiple adaptive thresholds and boundary evaluation［J］.IEEE Transactions on Intelligent Transportation Systems，2010，11(1):40-51.
[2]Dalal N，Triggs B.Histograms of oriented gradients for human detection［C］//Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition.San Diego:IEEE，2005:886-893.
[3]Cai Z，Fan Q，Feris R S，et al.A unified multi-scale deep convolutional neural network for fast object detection［C］ //European Conference on Computer Vision.Amsterdam:Springer，Cham，2016:354-370.
[4]Li J，Liang X，Shen S M，et al.Scale-aware fast R-CNN for pedestrian detection［J］.IEEE Transactions on Multimedia，2018，20(4):985-996.
[5]Du X，El-Khamy M，Morariu V I，et al.Fused deep neural networks for efficient pedestrian detection［J］.arXiv Preprint arXiv，2018:1805.08688.
[6]Brazil G，Yin X，Liu X.Illuminating pedestrians via simultaneous detection & segmentation［C］//Proceedings of IEEE International Conference on Computer Vision.Venice:IEEE，2017:4960-4969.
[7]Zhou C，Yang M，Yuan J.Discriminative feature transformation for occluded pedestrian detection［C］// Proceedings of IEEE/CVF International Conference on Computer Vision.Seoul:IEEE，2019:9556-9565.
[8]Zhang S，Yang J，Schiele B.Occluded pedestrian detection through guided attention in CNNs［C］//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Salt Lake City:IEEE，2018:6995-7003.
[9]Zhou C，Wu M，Lam S.SSA-CNN:semantic self-attention CNN for pedestrian detection［J］.arXiv Preprint arXiv，2019:1902.09080.
[10]Zhang L，Lin L，Liang X，et al.Is faster R-CNN doing well for pedestrian detection?［C］//European Conference on Computer Vision.Amsterdam:Springer，Cham，2016:443-457.
[11]Wu J，Zhou C，Yang M，et al.Temporal-context enhanced detection of heavily occluded pedestrians ［C］//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition.Seattle:IEEE，2020:13427-13436.
[12]Liu W，Anguelov D，Erhan D，et al.SSD:single shot multibox detector［C］//European Conference on Computer Vision.Amsterdam:Springer，Cham，2016:21-37.
[13]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 and Machine Intelligence，2017，39(6):1137-1149.
[14]Dollár 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.
[15]Simonyan K，Zisserman A.Very deep convolutional networks for large-scale image recognition［J］.arXiv Preprint arXiv，2014:1409-1556.
[16]Long J，Shelhamer E，Darrell T.Fully convolutional networks for semantic segmentation［C］//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Boston:IEEE，2015:3431-3440.
[17]Girshick R.Fast R-CNN［C］//Proceedings of the IEEE International Conference on Computer Vision.SanDiego:IEEE，2015:1440-1448.

基于语义分割注意力与可见区域预测的行人检测方法

Pedestrian Detection Based on Semantic Segmentation Attention and Visible Region Prediction

RichHTML

PDF (PC)

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]	杨丹，刘国如，任梦成，裴宏杨. 多尺度卷积核U-Net模型的视网膜血管分割方法[J]. 东北大学学报（自然科学版）, 2021, 42(1): 7-14.
[2]	原培新，陈鼎夫. 双能X射线高动态范围安检图像压缩算法[J]. 东北大学学报（自然科学版）, 2021, 42(1): 96-101.
[3]	冯宝，李昌林，李智，刘壮盛. 基于活动轮廓模型和影像组学的乳腺癌LVI状态预测[J]. 东北大学学报:自然科学版, 2020, 41(2): 193-199.
[4]	张云洲，郑瑞，暴吉宁，朱尚栋. 基于多个相关滤波器的行人跟踪尺度算法[J]. 东北大学学报:自然科学版, 2019, 40(9): 1228-1234.
[5]	陈红，于晓升，吴成东，孙鹏. 参数化水平集活动轮廓模型的快速图像分割算法[J]. 东北大学学报:自然科学版, 2019, 40(1): 6-10.
[6]	徐礼胜，张书琪，牛潇，徐阳. 基于全卷积网络的左心室射血分数自动检测[J]. 东北大学学报:自然科学版, 2018, 39(11): 1572-1576.
[7]	齐林，吕旭阳，杨本强，徐礼胜. 基于全卷积网络迁移学习的左心室内膜分割[J]. 东北大学学报:自然科学版, 2018, 39(11): 1577-1582.
[8]	魏国辉，齐守良，钱唯，张魁星. 基于相似性度量的肺结节图像检索算法[J]. 东北大学学报:自然科学版, 2018, 39(9): 1226-1231.
[9]	宫照煊，覃文军，郭薇，赵大哲. 改进的局部扩展拟合图像分割方法[J]. 东北大学学报:自然科学版, 2018, 39(4): 483-487.
[10]	王彬，赵海，朱宏博，朴春赫. 基于CT图像3D特征的肺结节检测[J]. 东北大学学报:自然科学版, 2018, 39(2): 181-185.
[11]	刘晓志，齐迪迪，贲驰. 基于畸变分离的摄像机标定方法[J]. 东北大学学报:自然科学版, 2017, 38(5): 620-624.
[12]	王琪，张铁，张晓梦，张祥德. 基于SIFT和SDM的虹膜定位方法[J]. 东北大学学报:自然科学版, 2017, 38(2): 180-184.
[13]	张祥德，朱和贵，李倩颖，唐青松. 基于MBC和POEM特征的人脸识别方法[J]. 东北大学学报:自然科学版, 2015, 36(11): 1526-1529.
[14]	彭怡书，颜云辉，赵久梁，张尧. 基于非局部稀疏特征的行人检测方法[J]. 东北大学学报:自然科学版, 2015, 36(4): 465-468.
[15]	杜小甫，王成恩. 基于等值线数据的一种新的云图算法[J]. 东北大学学报(自然科学版), 2013, 34(5): 624-627.