Abnormal crowd behavior detection using social force model

By Ramin Mehran, Alexis Oyama, Mubarak Shah. 2009 IEEE Conference on Computer Vision and Pattern Recognition, 2009

1,824 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Ocean Engineering

View Paper: https://www.semanticscholar.org/paper/c3113eaad326a955ba96c11b7b65d0c065fb2054

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Ocean Engineering

Citations in the corpus

You'll never walk alone: Modeling social behavior for multi-target tracking (2009, methodology): “[18] use the social force model to detect abnormal behavior in crowded scenes.”
Crowded Scene Analysis: A Survey (2015, methodology): “For that reason, several optical-ﬂow alike features have been presented in recent years [11]–[13], [22], [53]–[55], these methods avoid tracking from the macroscopic level and have achieved some success in addressing complex crowd ﬂows in the scenes.”
Anomaly Detection in Video via Self-Supervised and Multi-Task Learning (2020, methodology): “While some frameworks, such as [27, 33, 40, 41, 47], approach the problem from a global (frame-level) perspective, methods such as [7, 11, 21, 19, 28, 29, 31, 32, 44, 46, 64] extract features at a local level, e.”
Social Attention: Modeling Attention in Human Crowds (2017, background): “Subsequently, several approaches [11], [22] have extended the social forces model by fitting the parameters of the force functions to observed crowd behavior.”
Real-World Anomaly Detection in Surveillance Videos (2018, methodology): “Due to difficulties in obtaining reliable tracks, several approaches avoid tracking and learn global motion patterns through histogram-based methods [10], topic modeling [20], motion patterns [32], social force models [29], mixtures of dynamic textures model [27], Hidden Markov Model (HMM) on local ”
Summarizing surveillance videos with local-patch-learning-based abnormality detection, blob sequence optimization, and type-based synopsis (2015, background): “For example, Mehran [34] and Cui [35] derived social-force-based features from input optical flow fields to detect abnormal crowd activities.”
Chaotic invariants of Lagrangian particle trajectories for anomaly detection in crowded scenes (2010, methodology): “ROC curves for (a) our method, and (b) method of [9].”
A survey of video datasets for human action and activity recognition (2013, background): “UCF Web Dataset: Abnormal/Normal Crowds 2009 [128]/[129] Crowd Behavior”
Improving Data Association by Joint Modeling of Pedestrian Trajectories and Groupings (2010, methodology): “Applications include simulation [2–4], computer graphics [5, 6], and, in the last few years, also Computer Vision [7–14].”
Video-Based Abnormal Human Behavior Recognition—A Review (2012, methodology): “(i) UMN dataset: anomaly in crowd movement caused by panic [16].”
Social GAN: Socially Acceptable Trajectories with Generative Adversarial Networks (2018, methodology): “Over the past decades, this method has been often revisited [5, 6, 25, 26, 30, 31, 36, 46].”
Discriminative Latent Models for Recognizing Contextual Group Activities (2012, background): “For example, work has been done on exploiting contextual information between scenes and objects [4], objects and objects [5], [6], [7], objects and socalled “stuff” (amorphous spatial extent, e.g., trees, sky) [8], etc.”
Learning Deep Representations of Appearance and Motion for Anomalous Event Detection (2015, background): “To overcome the aforementioned limitations, researchers address the problem by learning spatio/temporal activity patterns in a local or a global context from either 2D image cells or 3D video volumes [2, 4, 8, 18, 21, 22, 24].”
What Do I Annotate Next? An Empirical Study of Active Learning for Action Localization (2018, background): “The computer vision community has embraced this problem, and during the last decade, several approaches for video analysis have been proposed [8,26,31,39,41,48,52,58,76].”
Detecting anomalous events in videos by learning deep representations of appearance and motion (2017, methodology): “[9] proposed a “social force” model based on optical flow features to represent crowd activity patterns and identify anomalous activities.”
Forecasting Interactive Dynamics of Pedestrians with Fictitious Play (2016, methodology): “In computer vision, the social force model has been used to help aid visual tracking [30] and anomaly detection [26].”
Abnormal Event Detection at 150 FPS in MATLAB (2013, methodology): “The videos are obtained from sources including dataset UCSD Ped1 [15], Subway datasets [1] (excluding abnormal event frames), 68 videos from YouTube, and 79 videos we captured.”
Social LSTM: Human Trajectory Prediction in Crowded Spaces (2016, methodology): “Neither, can it be predicted with simple ”repulsion” or ”attraction” functions (the traditional social forces models [24, 43, 73, 50]) This motivates us to build a model which can account for the behavior of other people within a large neighborhood, while predicting a person’s path.”