Analytics of deep neural network-based background subtraction

研究成果: ジャーナルへの寄稿記事

4 引用 (Scopus)

抄録

Deep neural network-based (DNN-based) background subtraction has demonstrated excellent performance for moving object detection. The DNN-based background subtraction automatically learns the background features from training images and outperforms conventional background modeling based on handcraft features. However, previous works fail to detail why DNNs work well for change detection. This discussion helps to understand the potential of DNNs in background subtraction and to improve DNNs. In this paper, we observe feature maps in all layers of a DNN used in our investigation directly. The DNN provides feature maps with the same resolution as that of the input image. These feature maps help to analyze DNN behaviors because feature maps and the input image can be simultaneously compared. Furthermore, we analyzed important filters for the detection accuracy by removing specific filters from the trained DNN. From the experiments, we found that the DNN consists of subtraction operations in convolutional layers and thresholding operations in bias layers and scene-specific filters are generated to suppress false positives from dynamic backgrounds. In addition, we discuss the characteristics and issues of the DNN based on our observation.

元の言語英語
記事番号78
ジャーナルJournal of Imaging
4
発行部数6
DOI
出版物ステータス出版済み - 1 1 2018

Fingerprint

Observation
Deep neural networks
Experiments

All Science Journal Classification (ASJC) codes

  • Radiology Nuclear Medicine and imaging
  • Computer Graphics and Computer-Aided Design
  • Computer Vision and Pattern Recognition
  • Electrical and Electronic Engineering

これを引用

@article{af748a2eb12e4ca5888a54a99739181d,
title = "Analytics of deep neural network-based background subtraction",
abstract = "Deep neural network-based (DNN-based) background subtraction has demonstrated excellent performance for moving object detection. The DNN-based background subtraction automatically learns the background features from training images and outperforms conventional background modeling based on handcraft features. However, previous works fail to detail why DNNs work well for change detection. This discussion helps to understand the potential of DNNs in background subtraction and to improve DNNs. In this paper, we observe feature maps in all layers of a DNN used in our investigation directly. The DNN provides feature maps with the same resolution as that of the input image. These feature maps help to analyze DNN behaviors because feature maps and the input image can be simultaneously compared. Furthermore, we analyzed important filters for the detection accuracy by removing specific filters from the trained DNN. From the experiments, we found that the DNN consists of subtraction operations in convolutional layers and thresholding operations in bias layers and scene-specific filters are generated to suppress false positives from dynamic backgrounds. In addition, we discuss the characteristics and issues of the DNN based on our observation.",
author = "Tsubasa Minematsu and Atsushi Shimada and Hideaki Uchiyama and Rin-Ichiro Taniguchi",
year = "2018",
month = "1",
day = "1",
doi = "10.3390/jimaging4060078",
language = "English",
volume = "4",
journal = "Journal of Imaging",
issn = "2313-433X",
number = "6",

}

TY - JOUR

T1 - Analytics of deep neural network-based background subtraction

AU - Minematsu, Tsubasa

AU - Shimada, Atsushi

AU - Uchiyama, Hideaki

AU - Taniguchi, Rin-Ichiro

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Deep neural network-based (DNN-based) background subtraction has demonstrated excellent performance for moving object detection. The DNN-based background subtraction automatically learns the background features from training images and outperforms conventional background modeling based on handcraft features. However, previous works fail to detail why DNNs work well for change detection. This discussion helps to understand the potential of DNNs in background subtraction and to improve DNNs. In this paper, we observe feature maps in all layers of a DNN used in our investigation directly. The DNN provides feature maps with the same resolution as that of the input image. These feature maps help to analyze DNN behaviors because feature maps and the input image can be simultaneously compared. Furthermore, we analyzed important filters for the detection accuracy by removing specific filters from the trained DNN. From the experiments, we found that the DNN consists of subtraction operations in convolutional layers and thresholding operations in bias layers and scene-specific filters are generated to suppress false positives from dynamic backgrounds. In addition, we discuss the characteristics and issues of the DNN based on our observation.

AB - Deep neural network-based (DNN-based) background subtraction has demonstrated excellent performance for moving object detection. The DNN-based background subtraction automatically learns the background features from training images and outperforms conventional background modeling based on handcraft features. However, previous works fail to detail why DNNs work well for change detection. This discussion helps to understand the potential of DNNs in background subtraction and to improve DNNs. In this paper, we observe feature maps in all layers of a DNN used in our investigation directly. The DNN provides feature maps with the same resolution as that of the input image. These feature maps help to analyze DNN behaviors because feature maps and the input image can be simultaneously compared. Furthermore, we analyzed important filters for the detection accuracy by removing specific filters from the trained DNN. From the experiments, we found that the DNN consists of subtraction operations in convolutional layers and thresholding operations in bias layers and scene-specific filters are generated to suppress false positives from dynamic backgrounds. In addition, we discuss the characteristics and issues of the DNN based on our observation.

UR - http://www.scopus.com/inward/record.url?scp=85063151364&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85063151364&partnerID=8YFLogxK

U2 - 10.3390/jimaging4060078

DO - 10.3390/jimaging4060078

M3 - Article

VL - 4

JO - Journal of Imaging

JF - Journal of Imaging

SN - 2313-433X

IS - 6

M1 - 78

ER -