期刊文献+

基于变分自编码器的视频异常事件检测方法 被引量:6

Video anomaly detection and localization via variational autoencoder
在线阅读 下载PDF
导出
摘要 异常事件检测由于其在视频监控场景中的重要性而引起了广泛的关注。但是由于缺乏异常标注样本,使得这个问题较难解决。提出了一种新的部分监督学习方法,仅采用正常样本训练检测模型以进行视频异常事件检测和定位。假设所有正常样本的分布符合一个高斯分布,那么异常样本在这个高斯分布中将以较低的概率出现。该方法基于变分自编码器(VAE),通过端对端的深度学习技术,将正常样本的隐层表示约束成一个高斯分布。给定测试样本,通过变分自编码器获得其隐层表示,计算其隐层表示属于高斯分布的概率,并根据检测门限判断其是否异常。在两个公开的数据集(UCSD dataset和avenue dataset)上的实验结果表明,所提出的方法达到了92.3%的帧级AUC和82.1%的帧级AUC,以及571 fps的检测速度,在性能和效率上明显高于现有检测方法。 Abnormal event detection has attracted wide attention due to its importance in video surveillance scenarios.The problem is particularly hard to crack because the lack of abnormal ground truth data.A novel partially supervised learning framework is presented for video anomaly detection and localization by training with the normal samples.The motivation is that the normal samples occur in high probability of a stochastic model,while the test samples occur in the low probability is regarded as anomaly.The method is based on variational auto-encoder(VAE),which can learn feature representations of the normal samples as a Gaussian distribution with deep learning technology.Experimental results of two popular benchmarks(UCSD dataset and avenue dataset)show that the proposed method achieved 92.3%and 82.1%frame-level AUC at a speed of 571 frames per second on average,which demonstrate the effectiveness and efficiency of the framework compared with other state of the arts approaches.
作者 苏鹏 王常顺 卢萌萌 Su Peng;Wang Changshun;Lu Mengmeng(Shandong Labor Vocational and Technical College,Jinan 250022,China;School of Information Science and Electrical Engineering,Shandong Jiaotong University,Jinan 250357,China)
出处 《电子测量与仪器学报》 CSCD 北大核心 2020年第10期179-185,共7页 Journal of Electronic Measurement and Instrumentation
基金 国家自然科学基金(61273144) 山东劳动职业技术学院校级科技项目(2019KJ02)资助。
关键词 视频监控 异常检测 深度学习 变分自编码器 video surveillance abnormal event detection deep learning variational auto-encoder
作者简介 苏鹏,分别在2008年和2011年于山东大学获得学士学位和硕士学位,现为山东劳动职业技术学院讲师、工程师,研究方向为控制工程、电气系统及其自动化。E-mail:treelucky_xl@163.com;王常顺,分别在2006年和2009年于山东大学获得学士学位和硕士学位,现为大连海事大学博士研究生,山东交通学院副教授,主要研究方向为AGV导航与智能控制无人艇导航与智能控制、嵌入式系统开发与应用。E-mail:wangchangshun@sdjtu.edu.cn;卢萌萌,2012年于山东轻工业学院获得学士学位,2015年于东华大学获得硕士学位,现为山东劳动职业技术学院讲师,主要研究方向为电力电子与电力传动技术。E-mail:lumengmeng@sdlvtc.cn。
  • 相关文献

参考文献4

二级参考文献80

  • 1钟志,徐扬生,石为人,叶伟中,李家强.群体异常检测(英文)[J].仪器仪表学报,2007,28(4):614-620. 被引量:4
  • 2POPOOLA O P, WANG K. Video-based abnormal hu- man behavior recognition-a review [ J ], IEEE Transac- tions on System, Man, and Cybernetics Part C, 2012,42 (6) : 865-878.
  • 3COLLINS R T, LIPTON A J, KANADE T. A system for video surveillance and monitoring [ C]. Proceedings of the 1999 American Nuclear Society (ANS) Eighth In- ternational Topical Meeting on Robotic and Remote Sys- tems, Pittsburgh, PA, USA ,25-29April, 1999.12-19.
  • 4HARITAOGLU I, HARWOOD D, DAVIS L S. W4 : A real time system for detecting and tracking people [ C ]. Proceedings of the 1998 IEEE International Conference on Computer Vision and Pattern Recognition, Santa Bar- bara, CA, USA, 23-25June, 1998,962-969.
  • 5HUANG K,TAN T. Vs-star: a visual interpretation sys- tem for visual surveillance[ J]. Pattern Recognition Let- ters,2010,31 ( 15 ) : 2265-2285.
  • 6University of California, San Diego. UCSD Anomaly De-tection Dataset [ EB/OL ]. http ://www. svcl. ucsd. edu/ project-s/anomaly/dataset, html. 2010-10-10.
  • 7University of Minnesota. UMN abnormal events detection dataset [ EB/OL ]. http ://mha. cs. umn. edu/proj _ e- vents, shtml. 2009-4-12.
  • 8ADAM A, RIVLIN E, SHIMSHONI I, REINITZ D. Ro- bust real-time unusual event detection using multiple fixed-location monitors [ J ]. IEEE Transactions on Pat- tern Analysis and Machine Intelligence, 2008,30 ( 3 ) : 555-560.
  • 9TAMRAKAR A, ALI S, YU Q, et al. Evaluation of low- level features and their combinations for complex event detection in open source videos [ C ]. Proceedings of the 2012 IEEE InlLernational Conference on Computer Vision and Pattern Recognition, Providence, RI, USA, 16- 21June,2012 : 3681-3688.
  • 10KWON J, LEE K M. A unified framework for event sum- marization and rare event detection [ C ]. Proceedings of the 2012 IEEE International Conference on Computer Vision and Pattern Recognition, Providence, RI, USA, 16-21June ,2012 : 1266-1273.

共引文献145

同被引文献50

引证文献6

二级引证文献46

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部