Abnormal detection using interaction energy potentials

Cui, Xinyi; Liu, Qingshan; Gao, Mingchen; Metaxas, Dimitris N.

doi:10.1109/cvpr.2011.5995558

Cited by 200 publications

(109 citation statements)

References 19 publications

Supporting

Mentioning

108

Contrasting

Unclassified

Order By: Relevance

“…5 shows the ROC curve of the abnormal behavior recognition on the UMN dataset for the proposed method using spatial+temporal descriptor. The performance of the proposed method is compared with state-of-the-art methods including BoTG [14], interaction energy potentials (IEP) [11], sparse reconstruction cost (SRC) [12], social force (SF) [8], streakline potential (SP) [10] and optical flow (OF). Table II …”

Section: Abnormal Behavior Recognitionmentioning

confidence: 99%

“…This gave a better representation of the flow in comparison with common flow representations, by more accurately recognizing the temporal and spatial changes in the scene. The use of potential function was also presented in [11] for abnormal crowd detection. An interaction energy potential function was designed to model the current behavior states of the subjects and their actions were represented by their velocities.…”

Section: Introductionmentioning

confidence: 99%

“…Object-based methods deal with a crowd as a collection of objects/individuals [2][3][4]. In holistic methods, a crowd is considered as a single entity and no segmentation or detection of individual objects is required [5][6][7][8][9][10][11][12][13][14]. Since the proposed method falls in the latter category, in the remainder of this section different holistic methods for crowd behavior analysis are discussed.…”

Section: Introductionmentioning

confidence: 99%

“…Crowd behavior understanding faces many challenges including group-level relationships, complex interactions, emergent behaviors and self-organizing activities [1]. To tackle these challenges, many techniques have been developed [2][3][4][5][6][7][8][9][10][11][12][13][14] which lead to various useful applications like crowd management, automatic surveillance, public space design and virtual and intelligent environments.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Crowd Behavior Recognition Using Dense Trajectories

Khokher

Bouzerdoum

Phung

2014

2014 International Conference on Digital Image Computing: Techniques and Applications (DICTA)

View full text Add to dashboard Cite

This article presents a new method for crowd behavior recognition, using dynamic features extracted from dense trajectories. The histogram of oriented gradient and motion boundary histogram descriptors are computed at dense points along motion trajectories, and tracked using median filtering and displacement information obtained from a dense optical flow field. Then a global representation of the scene is obtained using a bag-of-words model of the extracted features. The locality-constrained linear encoding with sum pooling and L2 plus power normalization are employed in the bag-of-words model. Finally, a support vector machine classifier is trained to recognize the crowd behavior in a short video sequence. The proposed method is tested on two benchmark datasets, and its performance is compared with those of some existing methods. Experimental results show that the proposed approach can achieve a classification rate of 93.8% on PETS2009 S3 and area under the curve score of 0.985 on UMN datasets respectively.

show abstract

Section: Abnormal Behavior Recognitionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Crowd Behavior Recognition Using Dense Trajectories

Khokher

Bouzerdoum

Phung

2014

2014 International Conference on Digital Image Computing: Techniques and Applications (DICTA)

View full text Add to dashboard Cite

show abstract

“…For example, J. Kim et al [1] uses the Markov Random Field (MRF) model and the maximum a posteriori (MAP) to estimate the abnormal degree, Y. Zhang [2] et al combines motion and appearance cues for anomaly detection based on Support Vector Data Description (SVDD). Totally, the abnormal events, including vehicle abnormal behavior [3,4], restricted-area access [5], group fighting [6] and carrying cases [7] can be detected.…”

Section: Introductionmentioning

confidence: 99%

New Approach of Abnormal Events Judgement and Early-warning Based on Combination of Audio Analysis and Video Analysis

Hu¹,

Shu²,

Tang³

et al. 2016

Proceedings of the 2016 International Conference on Computer Engineering and Information Systems

View full text Add to dashboard Cite

Abstract-The monitoring and early-warning of abnormal events in public places is an important part of the public security research. But nowadays, the monitoring way is basically video surveillance, and the judgement and analysis of abnormal events is based on the video content. This single way has low accuracy and analytic algorithms have some limitations, so the single audio analysis method can't offer timely and effective early-warning. Taking a sudden car fire event for example, this article combines the audio analysis and video analysis to judge abnormal events and also designs an new surveillance system, offering an effective way to early-warning.

show abstract