首页 > 论文 > 激光与光电子学进展 > 56卷 > 15期(pp:151001--1)

基于聚类式区域生成的全卷积目标检测网络

Full-Convolution Object Detection Network Based on Clustering Region Generation

  • 摘要
  • 论文信息
  • 参考文献
  • 被引情况
  • PDF全文
分享:

摘要

基于区域的全卷积网络(R-FCN)的区域生成网络(RPN)沿用了更快速区域卷积神经网络(Faster R-CNN)的RPN。针对RPN先验框的大小与数量均需人为固定,生成的建议区域过多等问题,将聚类思想应用到RPN中,改进先验框的生成方式,提出了基于聚类式区域生成的全卷积目标检测网络。通过对训练样本的真实框进行K-Means聚类得到先验框的最适大小和最佳数量,取代原本人为固定选取先验框的方式。此外,为增强模型的泛化能力,在改进后的R-FCN上使用ResNet基础网络,采用困难样本挖掘方法进行训练。实验结果表明,相较于R-FCN等方法,该聚类区域全卷积目标检测网络得到的检测结果在精度和速度上都得到了明显的提升。

Abstract

Region proposal networks (RPN) in region-based full-convolutional networks (R-FCN) follow the RPN of faster region convolutional neural networks. In this paper, a full-convolution object detection network based on clustering region generation is proposed to solve the problems of the artificially fixed sizes and quantities of anchor boxes and excessively generated proposals. K-means clustering on the ground-truth box of the training samples is used to optimize the sizes and numbers of the anchor boxes in order to replace the fixed boxes in the R-FCN. Furthermore, to enhance the generalization ability of the model, an online hard example mining is used to train the datasets based on the backbone network of ResNet. The experimental results show that the accuracy of the detection results of the proposed algorithm is significantly higher than that of the R-FCN.

Newport宣传-MKS新实验室计划
补充资料

DOI:10.3788/LOP56.151001

所属栏目:图像处理

基金项目:国家自然科学基金(61573168);

收稿日期:2019-01-04

修改稿日期:2019-03-05

网络出版日期:2019-08-01

作者单位    点击查看

潘志浩:江南大学物联网工程学院, 轻工过程先进控制教育部重点实验室, 江苏 无锡 214122
陈莹:江南大学物联网工程学院, 轻工过程先进控制教育部重点实验室, 江苏 无锡 214122

联系人作者:潘志浩(543152026@qq.com); 陈莹(chenying@jiangnan.edu.cn);

备注:国家自然科学基金(61573168);

【1】Ye G L, Sun S Y, Gao K J et al. Nighttime pedestrian detection based on faster region convolution neural network. Laser & Optoelectronics Progress. 54(8), (2017).
叶国林, 孙韶媛, 高凯珺 等. 基于加速区域卷积神经网络的夜间行人检测研究. 激光与光电子学进展. 54(8), (2017).

【2】Lu Y S, Li Y X, Liu B et al. Hyperspectral data haze monitoring based on deep residual network. Acta Optica Sinica. 37(11), (2017).
陆永帅, 李元祥, 刘波 等. 基于深度残差网络的高光谱遥感数据霾监测. 光学学报. 37(11), (2017).

【3】Wang Z L, Huang M, Zhu Q B et al. Smoke detection in storage yard based on parallel deep residual network. Laser & Optoelectronics Progress. 55(5), (2018).
王正来, 黄敏, 朱启兵 等. 基于并行深度残差网络的堆场烟雾检测方法. 激光与光电子学进展. 55(5), (2018).

【4】Girshick R, Donahue J, Darrell T et al. Region-based convolutional networks for accurate object detection and segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence. 38(1), 142-158(2016).

【5】He K M, Zhang X Y, Ren S Q et al. Spatial pyramid pooling in deep convolutional networks for visual recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence. 37(9), 1904-1916(2015).

【6】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 (CVPR), July 21-26, 2017, Honolulu, HI, USA. New York: IEEE. 17355379, (2017).

【7】Girshick R. Fast R-CNN. [C]∥2015 IEEE International Conference on Computer Vision (ICCV), December 7-13, 2015, Santiago, Chile. New York: IEEE. 1440-1448(2015).

【8】Ren S Q, He K M, Girshick R et al. Faster R-CNN: towards real-time object detection with region proposal networks. IEEE Transactions on Pattern Analysis and Machine Intelligence. 39(6), 1137-1149(2017).

【9】Dai J F, Li Y, He K M et al. R-FCN: object detection via region-based fully convolutional networks. [C]∥ 30th Conference on Neural Information Processing Systems (NIPS 2016), December 5-10, 2016, Barcelona, Spain. USA: Curran Associates Inc. 379-387(2016).

【10】He K M, Gkioxari G, Dollár P et al. Mask R-CNN. [C]∥2017 IEEE International Conference on Computer Vision (ICCV), October 22-29, 2017, Venice, Italy. New York: IEEE. 2980-2988(2017).

【11】Redmon J and Farhadi A. YOLO9000: better, faster, stronger. [C]∥2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), July 21-26, 2017, Honolulu, HI, USA. New York: IEEE. 6517-6525(2017).

【12】Redmon J and Farhadi A.Yolov3:an incremental improvement[J/OL]. -04-08)[2018-11-16]. https:∥arxiv. org/abs/1804, (2018).

【13】Liu W, Anguelov D, Erhan D et al. SSD: single shot MultiBox detector. ∥Leibe B, Matas J, Sebe N, et al. Computer vision-ECCV 2016. Lecture notes in computer science, Cham: Springer. 9905, 21-37(2016).

【14】Redmon J, Divvala S, Girshick R et al. You only look once: unified, real-time object detection. [C]∥2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 27-30, 2016, Las Vegas, NV, USA. New York: IEEE. 779-788(2016).

【15】Fu C Y, Liu W, Ranga A et al. -01-23)[2018-11-16]. https:∥arxiv. org/abs/1701, (2017).

【16】He K M, Zhang X Y, Ren S Q et al. Deep residual learning for image recognition. [C]∥2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 27-30, 2016, Las Vegas, NV, USA. New York: IEEE. 770-778(2016).

【17】Shrivastava A, Gupta A and Girshick R. Training region-based object detectors with online hard example mining. [C]∥2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 27-30, 2016, Las Vegas, NV, USA. New York: IEEE. 761-769(2016).

【18】Shrivastava M and Matlani P. A smoke detection algorithm based on K-means segmentation. [C]∥2016 International Conference on Audio, Language and Image Processing (ICALIP), July 11-12, 2016, Shanghai, China. New York: IEEE. 301-305(2016).

引用该论文

Pan Zhihao,Chen Ying. Full-Convolution Object Detection Network Based on Clustering Region Generation[J]. Laser & Optoelectronics Progress, 2019, 56(15): 151001

潘志浩,陈莹. 基于聚类式区域生成的全卷积目标检测网络[J]. 激光与光电子学进展, 2019, 56(15): 151001

您的浏览器不支持PDF插件,请使用最新的(Chrome/Fire Fox等)浏览器.或者您还可以点击此处下载该论文PDF