《计算机应用研究》|Application Research of Computers

基于自适应特征卷积网络的行人检测方法

Pedestrian detection method based on adaptive feature convolution network

免费全文下载 (已被下载 次)  
获取PDF全文
作者 陈乔松,弓攀豪,申发海,陶亚,董广县,王进,邓欣
机构 重庆邮电大学 数据工程与可视计算重点实验室,重庆 400065
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2020)07-059-2202-04
DOI 10.19734/j.issn.1001-3695.2019.02.0032
摘要 针对行人检测方法未能充分利用卷积网络浅层特征的问题,改进Faster R-CNN框架,提出了一种基于自适应特征卷积网络的行人检测方法。该方法有两处改进:a)设计了SFCM模块,用于提取卷积神经网络浅层细节特征;b)引用挤压与激励操作设计了AFCM模块,用于筛选检测所需的强辨识力行人特征。此外,利用公开的Caltech和INRIA行人数据集,通过在基准框架中逐一添加SFCM和AFCM模块训练行人检测器,验证了所提模块的有效性,并对比了主流行人检测算法。实验结果显示,所提方法的误检率分别降到了9.13%和9.46%,具有更优的检测性能。
关键词 行人检测; 卷积神经网络; 浅层细节特征; 自适应特征
基金项目 国家自然科学基金资助项目(61806033)
重庆市重点产业共性关键技术创新专项(cstc2017zdcy-zdyfX0012)
国家社会科学基金西部项目(18XGL013)
本文URL http://www.arocmag.com/article/01-2020-07-059.html
英文标题 Pedestrian detection method based on adaptive feature convolution network
作者英文名 Chen Qiaosong, Gong Panhao, Shen Fahai, Tao Ya, Dong Guangxian, Wang Jin, Deng Xin
机构英文名 Key Laboratory of Data Engineering & Visual Computing,Chongqing University of Posts & Telecommunications,Chongqing 400065,China
英文摘要 To circumvent the problem of failing to make full use of the shallow features of the convolutional network, this paper improved the existing Faster R-CNN framework and proposed pedestrian detection method based on adaptive feature convolution network, for achieving higher the detection accuracy. This paper has two improvements. Firstly, it designed SFCM module to extract the shallow detail features of the convolution neural network. Secondly, it proposed AFCM module by utilizing the squeeze and excitation mechanism, which was used to screen the strong discrimination features of pedestrian. Moreover, it used two public pedestrian datasets, Caltech and INRIA. It added SFCM module and AFCM module one by one in the benchmark framework, which verified the validity of the designed modules. Compared with some existing person detection algorithms, the experimental results show that the proposed method has better detection performance and miss rate dropped to 9.13% and 9.46% respectively.
英文关键词 pedestrian detection; convolution neural network; shallow detail features; adaptive feature
参考文献 查看稿件参考文献
 
收稿日期 2019/2/21
修回日期 2019/3/28
页码 2202-2205,2226
中图分类号 TP391.41
文献标志码 A