《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (2): 507-513.DOI: 10.11772/j.issn.1001-9081.2021122081

• 多媒体计算与计算机仿真 • 上一篇    

基于独立循环神经网络与变分自编码网络的视频帧异常检测

贾晴(), 王来花, 王伟胜   

  1. 曲阜师范大学 网络空间安全学院,山东 曲阜 273165
  • 收稿日期:2021-12-09 修回日期:2022-04-13 接受日期:2022-05-13 发布日期:2022-06-13 出版日期:2023-02-10
  • 通讯作者: 贾晴
  • 作者简介:王来花(1988—),女,山东聊城人,副教授,博士,主要研究方向:数字图像处理、视频异常检测
    王伟胜(1997—),男,山东烟台人,硕士研究生,主要研究方向:视频异常检测、计算机视觉。
  • 基金资助:
    国家自然科学基金资助项目(61601261)

Anomaly detection in video via independently recurrent neural network and variational autoencoder network

Qing JIA(), Laihua WANG, Weisheng WANG   

  1. School of Cyber Science and Engineering,Qufu Normal University,Qufu Shandong 273165,China
  • Received:2021-12-09 Revised:2022-04-13 Accepted:2022-05-13 Online:2022-06-13 Published:2023-02-10
  • Contact: Qing JIA
  • About author:WANG Laihua, born in 1988, Ph. D., associate professor. Her research interests include digital image processing, video anomaly detection.
    WANG Weisheng, born in 1997, M. S. candidate. His research interests include video anomaly detection, computer vision.
  • Supported by:
    National Natural Science Foundation of China(61601261)

摘要:

为了有效提取连续视频帧间的时间信息,提出一种融合独立循环神经网络(IndRNN)与变分自编码(VAE)网络的预测网络IndRNN-VAE。首先,利用VAE网络提取视频帧的空间信息,并通过线性变换得到视频帧的潜在特征;然后,将潜在特征作为IndRNN的输入以得到视频帧序列的时间信息;最后,通过残差块将获得的潜在变量与时间信息进行融合并输入到解码网络中来生成预测帧。通过在UCSD Ped1、UCSD Ped2、Avenue公开数据集上进行测试,实验结果表明,与现有的异常检测方法相比,基于IndRNN-VAE的方法性能得到了显著提升,曲线下面积(AUC)值分别达到了84.3%、96.2%和86.6%,错误率(EER)值分别达到了22.7%、8.8%和19.0%,平均异常得分的差值分别达到了0.263、0.497和0.293,且运行速度达到了每秒28帧。

关键词: 视频异常检测, 视频监控, 变分自编码器, 独立循环神经网络, 特征提取

Abstract:

To effectively extract the temporal information between consecutive video frames, a prediction network IndRNN-VAE (Independently Recurrent Neural Network-Variational AutoEncoder) that fuses Independently Recurrent Neural Network (IndRNN) and Variational AutoEncoder (VAE) network was proposed. Firstly, the spatial information of video frames was extracted through VAE network, and the latent features of video frames were obtained by a linear transformation. Secondly, the latent features were used as the input of IndRNN to obtain the temporal information of the sequence of video frames. Finally, the obtained latent features and temporal information were fused through residual block and input to the decoding network to generate the prediction frame. By testing on UCSD Ped1, UCSD Ped2 and Avenue public datasets, experimental results show that compared with the existing anomaly detection methods, the method based on IndRNN-VAE has the performance significantly improved, and has the Area Under Curve (AUC) values reached 84.3%, 96.2%, and 86.6% respectively, the Equal Error Rate (EER) values reached 22.7%, 8.8%, and 19.0% respectively, the difference values in the mean anomaly scores reached 0.263, 0.497, and 0.293 respectively. Besides, the running speed of this method reaches 28 FPS (Frames Per Socond).

Key words: video anomaly detection, video surveillance, Variational AutoEncoder (VAE), Independently Recurrent Neural Network (IndRNN), feature extraction

中图分类号: