Background subtraction in real applications: Challenges, current models and future directions

García-García, Belmar; Bouwmans, Thierry; Rosales, Alberto

doi:10.1016/j.cosrev.2019.100204

Cited by 241 publications

(110 citation statements)

References 94 publications

Supporting

Mentioning

109

Contrasting

Unclassified

Order By: Relevance

“…The deep learning method [30,53,54] has been widely used in computer vision in recent years. Unlike the most conventional vision-based fall detection methods [55] relying on hand-crafted features, the methods based on deep learning techniques can automatically learn features and hence have got widely concerned recently. Taramasco et al, in 2018, used a thermal sensor array for older people who live alone.…”

Section: Related Work and Contributionmentioning

confidence: 99%

Vision-Based Fall Event Detection in Complex Background Using Attention Guided Bi-Directional LSTM

Chen

Wang

et al. 2020

IEEE Access

View full text Add to dashboard Cite

Fall event, as one of the greatest risks to the elderly, its detection has been a hot research issue in the solitary scene in recent years. Nevertheless, most current researches are conducted in the ideal environments, without considering the challenge of complex background in real situation. Therefore, this paper aims to detect fall event detection in complex background based on visual data. Different from most conventional background subtraction methods which depend on background modeling, Mask R-CNN method is first used to accurately extract the moving objects in the noise background. Then, an attention guided Bi-directional LSTM model is proposed for the final fall event detection. To demonstrate the efficiency, the proposed method is verified in the public dataset and self-build dataset. Evaluation of the algorithm performances in comparison with other stateof-the-art methods indicates that the proposed design is accurate and robust, which means it is suitable for the task of fall event detection in complex situation.

show abstract

Section: Related Work and Contributionmentioning

confidence: 99%

Vision-Based Fall Event Detection in Complex Background Using Attention Guided Bi-Directional LSTM

Chen

Wang

et al. 2020

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Also, it is a challenging task to identify the objects with similar colour and patterns of the static background scenes of a video. Many works have been developed to cope this kind of difficulties, which initialize the background [1][2][3][4], subtracts background scenes [5][6][7], and forefront detection [8][9][10], that supports for video retrieval and detection of moving objects. The background subtraction, background initialization, foreground segmentation, and forefront object detection play a noteworthy role in video analysis like retrieval, summarization, classification, moving object detection and so on.…”

Section: Introductionmentioning

confidence: 99%

Key-Frame Detection and Video Retrieval Based on DC Coefficient-Based Cosine Orthogonality and Multivariate Statistical Tests

Kalakoti¹,

Prabakaran²

2020

View full text Add to dashboard Cite

This paper presents a method, which is developed based on the Discrete Cosine (DC) coefficient and multivariate parametric statistical tests, such as tests for equality of mean vectors and the covariance matrices. Background scenes and forefront objects are separated from the key-frame, and the salient features, such as colour and Gabor texture, are extracted from the background and forefront components. The extracted features are formulated as a feature vector. The feature vector is compared to that of the feature vector database, based on the statistical tests. First, the feature vectors are compared with respect to covariance. If the feature vector of the key-frame and the feature vector of the feature vector database pass the test, then the test for equality of mean vector is performed; otherwise, the testing process is stopped. If the feature vectors pass both tests, then it is inferred that the query key-frame represents the target video in the video database. Otherwise, it is concluded that the query key-frame not representing the video; and the proposed system takes the next feature vector for matching. The proposed method results in an average retrieval rate of 97.232%, 96.540%, and 96.641% for CC_WEB, UCF101, and our newly constructed database, respectively. Further, the mAP scores computed for each video datasets, which resulted in 0.807, 0.812, and 0.814 for CC_WEB, UCF101, and our newly constructed database, respectively. The output results obtained by the proposed method are comparable to the existing methods.

show abstract

“…Since considerable attention has been focused on background subtraction in recent years, promising methods have been proposed. Garcia-Garcia et al [8] presented an exhaustive review on background subtraction. Background subtrac-tion methods based on robust principal component analysis (RPCA) models [9], fuzzy models [10], semantic models [11] and deep learning models [12] are considered promising solutions.…”

Section: Introductionmentioning

confidence: 99%

Background Subtraction Based on GAN and Domain Adaptation for VHR Optical Remote Sensing Videos

Bai

Jiao

2020

IEEE Access

View full text Add to dashboard Cite

The application of deep learning techniques in background subtraction for VHR optical remote sensing videos holds the potential to facilitate multiple intelligent remote sensing processing tasks. However, existing methods on background subtraction for VHR optical remote sensing videos are still facing technical challenges. First, conventional CNN and other networks are limited by performance constraints. Second, existing background subtraction methods are mostly trained by natural videos due to the lack of VHR optical remote sensing video datasets. Third, VHR optical remote sensing videos have large scene sizes. In our article, we design a novel deep learning network via fully utilizing GAN and domain adaptation, which has the ability to measure and minimize the discrepancy between feature distributions of natural videos and VHR optical remote sensing videos so that the background subtraction performance for VHR optical remote sensing videos is improved significantly. Numerous experiments are conducted on the CDnet 2014 dataset and VHR optical remote sensing video dataset. Tremendous experiments demonstrate that our proposed method achieves an average FM of 0.8533, which reveals excellent performance on background subtraction. INDEX TERMS Background subtraction; Generative Adversarial Networks (GAN); domain adaptation; Very High Resolution (VHR) optical remote sensing videos

show abstract

Background subtraction in real applications: Challenges, current models and future directions

Cited by 241 publications

References 94 publications

Vision-Based Fall Event Detection in Complex Background Using Attention Guided Bi-Directional LSTM

Vision-Based Fall Event Detection in Complex Background Using Attention Guided Bi-Directional LSTM

Key-Frame Detection and Video Retrieval Based on DC Coefficient-Based Cosine Orthogonality and Multivariate Statistical Tests

Background Subtraction Based on GAN and Domain Adaptation for VHR Optical Remote Sensing Videos

Contact Info

Product

Resources

About