Motion pattern-based crowd scene classification using histogram of angular deviations of trajectories

Abstract

Automated crowd behaviour analysis and monitoring is a challenging task due to the unpredictable nature of the crowd within a particular scene and across different scenes. The prior knowledge of the type of scene under consideration is a crucial mid-level information, which could be utilized to develop robust crowd behaviour analysis systems. In this paper, we propose an approach to automatically detect the type of a crowded scene based on the global motion patterns of the objects within the scene. Three different types of scenes whose global motion pattern characteristics vary from uniform to non-uniform are considered in this work, namely structured, semi-structured, and unstructured scenes, respectively. To capture the global motion pattern characteristics of an input crowd scene, we first extract the motion information in the form of trajectories using a key-point tracker and then compute the average angular orientation feature of each trajectory. This paper utilizes these angular features to introduce a novel feature vector, termed as Histogram of Angular Deviations (HAD), which depicts the distribution of the pair-wise angular deviation values for each trajectory vector. Since angular deviation information is resistant to changes in scene perspectives, we consider it as a key feature for distinguishing the scene types. To evaluate the effectiveness of the proposed HAD-based feature vector in classifying the crowded scenes, we build a crowd scene classification model by training the classical machine learning algorithms on the publicly available Collective Motion Database. The experimental results demonstrate the superior crowd classification performance of the proposed approach as compared to the existing methods. In addition to this, we propose a technique based on quantizing the angular deviation values to reduce the feature dimension and subsequently introduce a novel crowd scene structuredness index to quantify the structuredness of an input crowded scene based on its HAD.

Motion Pattern-Based Scene Classification Using Adaptive Synthetic Oversampling and Fully Connected Deep Neural Network

“…Despite the abundance of research conducted on motion pattern-based crowd analysis [3], [9], [10], there exists only a limited body of work specifically targeting the classification of a scene into the three distinct categories of interest: structured, semi-structured, and unstructured [7], [11]. Most existing research has explored various aspects of motion patterns without delving into the precise categorization that distinguishes between these three fundamental types of crowd behavior.…”

“…Subsequently, histograms of these motion orientations at the block level are combined to form comprehensive frame-level features. In [7], a feature vector was introduced that utilizes the histogram of angular deviations from average trajectory vectors to categorize crowded scenes as either structured, semi-structured, or unstructured, depending on overarching motion patterns. The methodology put forth comprises values that represent the frequency of each conceivable angular deviation, ranging from 0 degrees to 180 degrees.…”

“…Similar to [7] and [11], many of the previously discussed methods calculate a collectiveness value for each individual frame and then take an average across all frames to determine the overall crowd collectiveness of a given video. Such perframe methodologies tend to yield considerable fluctuations in the measured collectiveness, owing mainly to the everchanging motion patterns of the trajectory's key points from one frame to the next.…”

“…Our method fundamentally diverges from [7] and [11], as it introduces a refined, fully connected neural network model aimed at segregating crowds into three well-defined categories: structured, semi-structured, and unstructured. This classification hinges on the consistency or randomness of the movement behaviors displayed by the assembled entities.…”

“…Crowd density estimation and crowd counting [5] are focused on assessing the number of people and the compactness of a crowd, which have applications in public safety and event management. Crowd anomaly detection [6] identifies unusual patterns that may indicate danger or suspicious activities, while group detection [7] explores the formation and behavior of smaller groups within the crowd. These research areas contribute to developing sophisticated monitoring and management systems with broad applications in urban planning, law enforcement, and emergency response.…”