CASIA OpenIR  > 模式识别国家重点实验室  > 多媒体计算与图形学
基于改进RPN深度网络的端到端的监控场景行人检测研究
杨东明1,2; 徐士彪3; 孟维亮3; 葛水英1; 杨真1,2; 张晓鹏3
Source Publication中国体视学与图像分析
2017
Volume22Issue:2Pages:209-215
Other Abstract Compared to other scenes, pedestrian detection in monitoring scenes gets larger flow of people and higher degree of occlusion. This paper presents an end - to - end detection scheme based on region proposal network ( RPN). On the one hand, we improved RPN by combining with a deep convolution network designed by ourselves to obtain a new deep neural network for pedestrian detection. On the other hand, we introduced head - shoulders model to improved detection performance for pedestrian detection at monitoring scenes, which improves the detection speed. Eventually, we achieved end - to - end detection. The experimental results show that the method effectively improved the detection performance by reducing the missing rate and speeding up the detection.
Keyword监控 行人检测 深度学习 Rpn网络
Document Type期刊论文
Identifierhttp://ir.ia.ac.cn/handle/173211/15674
Collection模式识别国家重点实验室_多媒体计算与图形学
Affiliation1.中国科学院文献情报中心
2.中国科学院大学
3.中国科学院自动化研究所
Recommended Citation
GB/T 7714
杨东明,徐士彪,孟维亮,等. 基于改进RPN深度网络的端到端的监控场景行人检测研究[J]. 中国体视学与图像分析,2017,22(2):209-215.
APA 杨东明,徐士彪,孟维亮,葛水英,杨真,&张晓鹏.(2017).基于改进RPN深度网络的端到端的监控场景行人检测研究.中国体视学与图像分析,22(2),209-215.
MLA 杨东明,et al."基于改进RPN深度网络的端到端的监控场景行人检测研究".中国体视学与图像分析 22.2(2017):209-215.
Files in This Item: Download All
File Name/Size DocType Version Access License
06体视学_基于改进RPN深度网络的端到(2115KB)期刊论文作者接受稿开放获取CC BY-NC-SAView Download
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[杨东明]'s Articles
[徐士彪]'s Articles
[孟维亮]'s Articles
Baidu academic
Similar articles in Baidu academic
[杨东明]'s Articles
[徐士彪]'s Articles
[孟维亮]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[杨东明]'s Articles
[徐士彪]'s Articles
[孟维亮]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: 06体视学_基于改进RPN深度网络的端到端的监控场景行人检测研究_杨东明.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.