计算机科学 ›› 2021, Vol. 48 ›› Issue (12): 264-268.doi: 10.11896/jsjkx.201200196

• 计算机图形学&多媒体 • 上一篇    下一篇

基于相邻特征融合的目标检测

李亚泽, 刘宏哲   

  1. 北京联合大学北京市信息服务工程重点实验室 北京100101
    北京联合大学机器人学院 北京100101
  • 收稿日期:2020-12-22 修回日期:2021-06-08 出版日期:2021-12-15 发布日期:2021-11-26
  • 通讯作者: 刘宏哲(liuhongzhe@buu.edu.cn)
  • 作者简介:yaze_li@126.com
  • 基金资助:
    国家自然科学基金(61871039,61906017,61802019);北京市教委项目(KM202111417001,KM201911417001;视觉智能协同创新中心项目(CYXC2011);北京联合大学学术项目(ZK80202001,202011417004,202011417005)

Object Detection Based on Neighbour Feature Fusion

LI Ya-ze, LIU Hong-zhe   

  1. Beijing Key Laboratory of Information Service Engineering,Beijing Union University,Beijing 100101,China
    College of Robotics,Beijing Union University,Beijing 100101,China
  • Received:2020-12-22 Revised:2021-06-08 Online:2021-12-15 Published:2021-11-26
  • About author:LI Ya-ze,born in 1991,postgraduate.His main research interests include computer vision and object detection.
    LIU Hong-zhe,born in 1971,Ph.D.Her main research interests include compu-ter vision,deep learning,media semantic computing,etc.
  • Supported by:
    National Natural Science Foundation of China(61871039,61906017,61802019),Beijing Municipal Commission of Education Project(KM202111417001,KM201911417001),Collaborative Innovation Center for Visual Intelligence(CYXC2011) and Academic Research Projects of Beijing Union University(ZK80202001,202011417004,202011417005).

摘要: 随着智能驾驶领域的发展,人们对目标检测的精度要求越来越高,尤其是针对高速行驶时对距离较远的小目标的检测和低速行驶时对密集目标的检测。在当前的两阶段检测框架的特征融合部分,使用bottom-up的双向融合方法虽然能够更有效地对大目标进行语义信息和位置信息的特征融合,但会给几个或几十个像素的小目标造成很大的信息损失。当检测网络特征融合部分使用top-down的单向融合方法时,则对大目标检测的效果欠佳。为此,文中提出了相邻特征融合(Neighbour Feature Pyramid Network,NFPN)方法、Double RoI(Region of Interest)方法和递归特征金字塔(Recursive Feature Pyramid,RFP)的方法。以Faster RCNN 50为基准,同时使用提出的NFPN,Double RoI和RFP后,在Lisa交通数据集中平均精度(mAP)提升了2.6个百分点。在VOC2007数据集上,以VOC07+12 train数据集为训练集,VOC2007 test为测试集,以Faster RCNN101为基准,同时使用提出的3个模型,mAP提升了6个百分点,同时小、中、大目标的精度也得到提高。

关键词: 深度学习, 目标检测, 计算机视觉, 特征融合, 智能驾驶

Abstract: With the development of intelligent driving,the precision requirements for target detection are getting higher and higher,especially for small targets that are far away.In the neck of two-stage object detection network,although the feature fusion of semantic information and location information is more effective for large targets if the bottom-up fusion method is used,it will cause big information loss to small targets.To address this problem,we propose neighbor feature pyramid networks(NFPN) method of feature fusion of neighbor layers,the Double RoI(Region of Interest) method to fuse the FPN and NFPN features,and the recursive feature pyramicl(RFP) method.Using Faster RCNN 50 as the benchmark,the mean average precision(mAP) of our model in the Lisa data set has increased by 2.6% while using NFPN,Double RoI and RFP.On the VOC2007 data set,using the VOC07+12 train data set for training,VOC2007 test as the test set,and Faster RCNN101 as the baseline,the mAP of our model both used NFPN,Double RoIE and RFP has increased by 6%,and the object detect accuracy of large,medium and small targets is improved at the same time.

Key words: Deep learning, Object detection, Computer vision, Feature fusion, Autonomous driving

中图分类号: 

  • TP183
[1]MOGELMOSE A,TRIVEDI M M,MOESLUND T B.Vision-based traffic sign detection and analysis for intelligent driver assistance systems:Perspectives and survey[J].IEEE Transactions on Intelligent Transportation Systems,2012,13(4):1484-1497.
[2]HE K,ZHANG X,REN S,et al.Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2016:770-778.
[3]REN S,HE K,GIRSHICK R,et al.Faster r-cnn:Towards real-time object detection with region proposal networks[C]//Advances in Neural Information Processing Systems.2015:91-99.
[4]LIU S,QI L,QIN H,et al.Path aggregation network for instance segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2018:8759-8768.
[5]LIN T Y,DOLLÁR P,GIRSHICK R,et al.Feature pyramid networks for object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2017:2117-2125.
[6]JOHN M E.The PASCAL Visual Object Classes Challenge 2007(VOC2007) Development Kit[J].International Journal of Computer Vision,2006,111(1):98-136.
[7]LIU W,ANGUELOV D,ERHAN D,et al.Ssd:Single shot multibox detector[C]//European Conference on Computer Vision.2016:21-37.
[8]TAN M,PANG R,LE Q V.Efficientdet:Scalable and efficient object detection[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2020:10781-10790.
[9]PANG J,CHEN K,SHI J,et al.Libra r-cnn:Towards balanced learning for object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2019:821-830.
[10]HE K,GKIOXARI G,DOLLÁR P,et al.Mask r-cnn[C]//Pro- ceedings of the IEEE International Conference on Computer Vision.2017:2961-2969.
[11]REDMON J,FARHADI A.Yolov3:An incremental improve- ment[J].arXiv:1804.02767,2018.
[12]LIN T Y,GOYAL P,GIRSHICK R,et al.Focal loss for dense object detection[C]//Proceedings of the IEEE International Conference on Computer Vision.2017:2980-2988.
[13]CAI Z,VASCONCELOS N.Cascade r-cnn:Delving into high quality object detection[C]//Proceedings of the IEEE Confe-rence on Computer Vision and Pattern Recognition.2018:6154-6162.
[14]QIAO S,CHEN L C,YUILLE A.DetectoRS:Detecting Objects with Recursive Feature Pyramid and Switchable Atrous Convolution[J].arXiv:2006.02334,2020.
[15]LIU Y,WANG Y,WANG S,et al.CBNet:A Novel Composite Backbone Network Architecture for Object Detection[C]//AAAI.2020:11653-11660.
[16]SHRIVASTAVA A,GUPTA A,GIRSHICK R.Training region-based object detectors with online hard example mining[C]//IEEE Conference on Computer Vision & Pattern Recognition.IEEE Computer Society,2016:761-769.
[17]CHEN K,WANG J,PANG J,et al.Mmdetection:Open mmlab detection toolbox and benchmark[J].arXiv:1906.07155,2019.
[18]GAO S,CHENG M M,ZHAO K,et al.Res2Net:A New Multi-scale Backbone Architecture[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2021,43(2):652-662,1.
[19]WANG T,YUAN L,ZHANG X,et al.Distilling object detectors with fine-grained feature imitation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2019:4933-4942.
[1] 董晓梅, 王蕊, 邹欣开. 面向推荐应用的差分隐私方案综述[J]. 计算机科学, 2021, 48(9): 21-35.
[2] 周新民, 胡宜桂, 刘文洁, 孙荣俊. 基于多模态多层级数据融合方法的城市功能识别研究[J]. 计算机科学, 2021, 48(9): 50-58.
[3] 钱梦薇, 过弋. 融合偏置深度学习的距离分解Top-N推荐算法[J]. 计算机科学, 2021, 48(9): 103-109.
[4] 徐涛, 田崇阳, 刘才华. 基于深度学习的人群异常行为检测综述[J]. 计算机科学, 2021, 48(9): 125-134.
[5] 赫晓慧, 邱芳冰, 程淅杰, 田智慧, 周广胜. 基于边缘特征融合的高分影像建筑物目标检测[J]. 计算机科学, 2021, 48(9): 140-145.
[6] 张新峰, 宋博. 一种基于改进三元组损失和特征融合的行人重识别方法[J]. 计算机科学, 2021, 48(9): 146-152.
[7] 袁磊, 刘紫燕, 朱明成, 马珊珊, 陈霖周廷. 融合改进密集连接和分布排序损失的遥感图像检测[J]. 计算机科学, 2021, 48(9): 168-173.
[8] 林椹尠, 张梦凯, 吴成茂, 郑兴宁. 利用生成对抗网络的人脸图像分步补全法[J]. 计算机科学, 2021, 48(9): 174-180.
[9] 黄晓生, 徐静. 基于PCANet的非下采样剪切波域多聚焦图像融合[J]. 计算机科学, 2021, 48(9): 181-186.
[10] 田野, 陈宏巍, 王法胜, 陈兴文. 室内移动机器人的SLAM算法综述[J]. 计算机科学, 2021, 48(9): 223-234.
[11] 谢良旭, 李峰, 谢建平, 许晓军. 基于融合神经网络模型的药物分子性质预测[J]. 计算机科学, 2021, 48(9): 251-256.
[12] 冯霞, 胡志毅, 刘才华. 跨模态检索研究进展综述[J]. 计算机科学, 2021, 48(8): 13-23.
[13] 王立梅, 朱旭光, 汪德嘉, 张勇, 邢春晓. 基于深度学习的民事案件判决结果分类方法研究[J]. 计算机科学, 2021, 48(8): 80-85.
[14] 龚浩田, 张萌. 基于关键点检测的无锚框轻量级目标检测算法[J]. 计算机科学, 2021, 48(8): 106-110.
[15] 叶中玉, 吴梦麟. 融合时序监督和注意力机制的脉络膜新生血管分割[J]. 计算机科学, 2021, 48(8): 118-124.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
[1] 周丹晨. 融合粗糙集和商空间的企业级信息系统日志挖掘方法[J]. 计算机科学, 2014, 41(Z6): 421 -424 .
[2] 丁勇,朱辉生,曹红根. 基于混合EHMM模型的数据流预测[J]. 计算机科学, 2014, 41(Z6): 391 -393 .
[3] 李修云. 基于Activiti框架的在线审批流程应用研究[J]. 计算机科学, 2016, 43(Z6): 555 -557 .
[4] 杨洁, 王国胤, 张清华, 冯林. 层次粒结构下粗糙模糊集的不确定性度量[J]. 计算机科学, 2019, 46(1): 45 -50 .
[5] 刘丽倩, 董东. 基于代价敏感集成分类器的长方法检测[J]. 计算机科学, 2018, 45(11A): 497 -500 .
[6] 李建军, 侯跃, 杨玉. 基于情景感知的用户兴趣推荐模型[J]. 计算机科学, 2019, 46(6A): 502 -506 .
[7] 程盛淦, 于浩然, 韦建文, 林新华. 基于定点压缩技术的双层粒子网格算法的设计与优化[J]. 计算机科学, 2020, 47(8): 56 -61 .
[8] 杨如涵, 戴毅茹, 王坚, 董津. 基于表示学习的工业领域人机物本体融合[J]. 计算机科学, 2021, 48(5): 190 -196 .
[9] 潘孝勤, 芦天亮, 杜彦辉, 仝鑫. 基于深度学习的语音合成与转换技术综述[J]. 计算机科学, 2021, 48(8): 200 -208 .
[10] 王俊, 王修来, 庞威, 赵鸿飞. 面向科技前瞻预测的大数据治理研究[J]. 计算机科学, 2021, 48(9): 36 -42 .