基于深度学习的行人检测方法在图书馆中的应用研究

图书馆杂志 ›› 2021, Vol. 40 ›› Issue (9): 62-69.

基于深度学习的行人检测方法在图书馆中的应用研究

牛悦李辉刘钊

出版日期:2021-09-15 发布日期:2021-09-27
作者简介:牛悦女，西北工业大学图书馆信息咨询服务部，馆员。研究方向：智慧数字图书馆构建、用户教育与培训、新媒体。作者贡献：文章整体架构、行人检测技术算法实现、实验搭建、实验结果分析、文章撰写等。E-mail：sjxz@nwpu.edu.cn 陕西西安 710072 李辉西北工业大学图书馆信息咨询服务部，副馆长，副研究员。研究方向：图书馆学、出版学、公共管理及国际关系。作者贡献：行人检测技术和图书馆具体应用的结合分析。陕西西安 710072 刘钊西北工业大学图书馆信息咨询服务部，馆员。研究方向：学科服务建设和智库建设。作者贡献：行人检测技术数据集收集和实验运行。陕西西安 710072

Application of Deep Learning Based on Pedestrian Detection in Library

Niu Yue, Li Hui, Liu Zhao

Online:2021-09-15 Published:2021-09-27
About author:Niu Yue, Li Hui, Liu Zhao (Library, Northwestern Polytechnical University )

摘要/Abstract

摘要： 深度学习和计算机视觉技术在产业升级中发挥了重要的价值，是建设下一代智慧图书馆的关键技术，而行人检测技术是深度学习结合计算机视觉技术重要的研究应用方向。本文探索了行人检测技术在图书馆中的应用，探讨了该技术对图书馆服务的重要意义，并且成功地将此项技术的代表方法YOLOv3实践在西北工业大学图书馆场景中，详细讨论了行人检测技术如何提高西北工业大学图书馆服务的应用实例。我们在行人数据库上训练了YOLOv3网络模型，并将其部署在西北工业大学的图书馆中，代表性场景的行人检测结果表明，YOLOv3在精度和实时性上能够满足图书馆场景下行人检测的需求。本文是将深度学习和计算机视觉技术用于构建智慧图书馆的一次有益尝试，也为其他相关技术在图书馆的落地实践提供了较好的借鉴作用。

Abstract: Deep learning and computer vision play an important role and is of great value in industrial upgrading; they are key technologies for constructing next-generation intelligent library. Pedestrian detection is an important research and application direction of this field. This work explores the application of pedestrian detection method to library for the first time, and discusses the important value
of the related applications in detail based on some real examples. The authors create the pedestrian detection representative method – YOLOv3 on open pedestrian databases, and successfully deploy the detection system in NWPU library. Some representative scenarios show that YOLOv3 can satisfy library needs both in terms of accuracy and of run-time performance. This work is a valuable attempt of applying
deep learning and computer vision technology in smart library field, and has good reference for utilizing related technologies in library.

牛悦李辉刘钊. 基于深度学习的行人检测方法在图书馆中的应用研究[J]. 图书馆杂志, 2021, 40(9): 62-69.

Niu Yue, Li Hui, Liu Zhao. Application of Deep Learning Based on Pedestrian Detection in Library[J]. Libraly Journal, 2021, 40(9): 62-69.

参考文献

[ 1 ] 初景利, 段美珍. 智慧图书馆与智慧服务[J]. 图书
馆建设, 2018(4): 85-90; 95.
[ 2 ] 储节旺, 陈梦蕾. 人工智能驱动图书馆变革[J]. 大
学图书馆学报, 2019, 37(4): 5-13.
[ 3 ] 姚飞, 纪磊, 张成昱, 等. 实时虚拟参考咨询服务
新尝试——清华大学图书馆智能聊天机器人[J].
现代图书情报技术, 2011(4): 77-81.
[ 4 ] 何秀美, 朱庆华, 沈超. 用户场景驱动的移动视
觉搜索资源组织研究[J]. 图书馆学研究, 2019(4):
46-52.
[ 5 ] 魏清华, 孙林, 胡文静. 高校图书馆特藏文献数字
化建设研究——以CASHL为例[J]. 大学图书情报
学刊, 2020, 38(1): 81-86.
[ 6 ] R Girshick, J Donahue, T Darrell, et al. Rich Feature
Hierarchies for Accurate Object Detection and
Semantic Segmentation. 2014 IEEE Conference on
Computer Vision and Pattern Recognition[EB/OL].
Columbus, 2014: 580-587.
[ 7 ] 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.
[ 8 ] Liu Wei, Anguelov, Dragomir, et al. SSD: Single Shot
Multi Box Detector. In Computer Vision - ECCV
2016[EB/OL]. Lecture Notes in Computer Science,
2016, 9905 (Springer).
[ 9 ] J Redmon, S Divvala, R Girshick, et al. You Only Look
Once: Unified, Real-Time Object Detection. 2016
IEEE Conference on Computer Vision and Pattern
Recognition (CVPR) [EB/OL]. 2016: 779-788.
[10] Redmon, Joseph & Farhadi, Ali. YOLOv3: An
Incremental Improvement, arXiv, 1804.02767[EB/
OL]. 2018.
[11] Fu Chengyang, Liu wei, A Ranga, et al. DSSD:
Deconvolutional single shot detector. arXiv:
1701.06659[EB/OL]. [2017- 01-23].
[12] T Y Lin, P Goyal, R Girshick, et al. Focal loss for
dense object detection[J/OL]. IEEE Transactions on
Pattern Analysis and Machine Intelligence IEEE
Trans, 2020, 42(2): 318-327.
[13] G Overett, L Petersson, N Brewer L, et al. A new
pedestrian dataset for supervised learning[EB/OL].
Intelligent Vehicles Symposium, 2008.
[14] N Dalal, B Triggs. Histograms of oriented gradients
for human detection[EB/OL]. IEEE Computer
Society Conference on Computer Vision and Pattern
Recognition, 2005, 1: 886-893.
[15] 董同强, 唐晓旺. 5G+AI: 构建“智能+”时代
的新一代图书馆服务平台[J]. 图书馆学研究,
2020(5): 81-86.