A low dimensional descriptor for detection of anomalies in crowd videos
基于低维度描述子的人群视频异常行为检测
Mathematics and Computers in Simulation, In press, corrected proof, Available online 3 June 2019
Tehreem Qasim, Naeem Bhatti
摘要：In this paper a novel descriptor is proposed for anomaly detection in crowd videos at a global level. Traditional approaches for anomaly detection in crowd videos face the dilemma of trade-off between high accuracy and real time performance. In order to resolve this issue, we propose an efficient descriptor composed of three different features extracted from the optical flow (OF) of a video sequence. The first feature is the sum of the optical flow field magnitude computed after applying a threshold. The second feature is the joint entropy of the OF magnitude of two consecutive frames used to measure the dissimilarity. The third feature is the variance computed from a space–time cuboid constructed using history of the OF field magnitude. Performance of the proposed descriptor is evaluated on the widely used UMN dataset in terms of accuracy and processing time. For UMN dataset, the proposed descriptor provides the highest area under the curve (AUC) compared to the approaches already published in literature.
EDCAR: A knowledge representation framework to enhance automatic video surveillance
EDCAR: 增强自动视频监控的知识表示框架
Expert Systems with Applications, Volume 131, 1 October 2019, Pages 190-207
Loredana Caruccio, Giuseppe Polese, Genoveffa Tortora, Daniele Iannone
摘要：The main purpose of video-based event recognition is to interpret activities or behaviors within video sequences, in order to detect and isolate specific events, which have to be readily recognized and prompted to the people responsible for their monitoring. In this paper, we present a knowledge representation framework and a system for automatic video surveillance, which analyzes record scenes in order to detect the occurrence of specific events defined as targets. The framework, named Elements and Descriptors of Context and Action Representations (EDCAR), enables the representation of relevant elements, general descriptors of the context, and actions that have to be captured, including the definition of action compositions and sequences, in order to monitor and recognize abnormal situations. EDCAR and the associated system also support video summarization of relevant scenes, providing an inference engine to handle complex queries. They have been used experimentally on several video surveillance scenarios, which enabled us to prove their effectiveness with respect to similar solutions described in the literature.
Using the Ensemble of Deep Neural Networks for Normal and Abnormal Situations Detection and Recognition in the Continuous Video Stream of the Security System
基于深度神经网络集成的安全系统连续视频流中的正常与异常状况检测与识别
Procedia Computer Science, Volume 150, 2019, Pages 532-539
O. S. Amosov, S. G. Amosova, Y. S. Ivanov, S. V. Zhiganov
摘要：It is suggested to use the ensemble of deep neural networks to design an intellectual situation classifier that solves the problem of normal and abnormal situations detection and recognition in a continuous video stream of the security system. The testing of its work was made on the basis of modern hardware and software and computer vision methods, the result of which is the classification probabilities for each video fragment. A software module in Python was created for normal and abnormal situations detection and recognition.
AED-Net: An Abnormal Event Detection Network
AED-Net: 异常事件检测网络
Engineering, In press, accepted manuscript, Available online 25 May 2019
Tian Wang, Zichen Miao, Yuxin Chen, Yi Zhou, Hichem Snoussi
摘要：It has long been a challenging task to detect an anomaly in a crowded scene. In this paper, a self-supervised framework called the abnormal event detection network (AED-Net), which is composed of a principal component analysis network (PCAnet) and kernel principal component analysis (kPCA), is proposed to address this problem. Using surveillance video sequences of different scenes as raw data, the PCAnet is trained to extract high-level semantics of the crowd’s situation. Next, kPCA, a one-class classifier, is trained to identify anomalies within the scene. In contrast to some prevailing deep learning methods, this framework is completely self-supervised because it utilizes only video sequences of a normal situation. Experiments in global and local abnormal event detection are carried out on Monitoring Human Activity dataset from University of Minnesota (UMN dataset) and Anomaly Detection dataset from University of California, San Diego (UCSD dataset), and competitive results that yield a better equal error rate (EER) and area under curve (AUC) than other state-of-the-art methods are observed. Furthermore, by adding a local response normalization (LRN) layer, we propose an improvement to the original AED-Net. The results demonstrate that this proposed version performs better by promoting the framework’s generalization capacity.
