A deep crowd density classification model for Hajj pilgrimage using fully convolutional neural network

Abstract

This research enhances crowd analysis by focusing on excessive crowd analysis and crowd density predictions for Hajj and Umrah pilgrimages. Crowd analysis usually analyzes the number of objects within an image or a frame in the videos and is regularly solved by estimating the density generated from the object location annotations. However, it suffers from low accuracy when the crowd is far away from the surveillance camera. This research proposes an approach to overcome the problem of estimating crowd density taken by a surveillance camera at a distance. The proposed approach employs a fully convolutional neural network (FCNN)-based method to monitor crowd analysis, especially for the classification of crowd density. This study aims to address the current technological challenges faced in video analysis in a scenario where the movement of large numbers of pilgrims with densities ranging between 7 and 8 per square meter. To address this challenge, this study aims to develop a new dataset based on the Hajj pilgrimage scenario. To validate the proposed method, the proposed model is compared with existing models using existing datasets. The proposed FCNN based method achieved a final accuracy of 100%, 98%, and 98.16% on the proposed dataset, the UCSD dataset, and the JHU-CROWD dataset, respectively. Additionally, The ResNet based method obtained final accuracy of 97%, 89%, and 97% for the proposed dataset, UCSD dataset, and JHU-CROWD dataset, respectively. The proposed Hajj-Crowd-2021 crowd analysis dataset and the model outperformed the other state-of-the-art datasets and models in most cases.

Crowd Management Intelligence Framework: Umrah Use Case

“…Over the years, significant research efforts have been dedicated to developing sophisticated solutions and strategies for effective crowd management. Crowd management strategies include Crowd counting [12], [13], to quantify the number of individuals in a specific area; Density estimation [14]- [16], to estimate crowd density in a given area; Localization and tracking [17]- [20], to locate and track individuals or groups within a crowd of people; Behavior monitoring [21]- [24], to analyze and interpret individuals' actions within a crowd; and Anomaly detection [25]- [35], to identify unusual or unexpected events within the crowd.…”

“…Crowd management strategies mostly rely on camera-based data sources [12], [14], [16], [17], [21], [36] or sensor-based data sources [25], [37], while some innovative approaches combine camera and sensor data to improve crowd management outcomes [19], [22]. Studies in crowd management adopt two methodologies: simulation and modeling.…”

“…These simulations facilitate the creation of different crowd scenarios, allowing an in-depth evaluation of diverse management strategies. Alternatively, other researchers have developed models leveraging deep learning (DL) techniques to detect or predict crowd patterns [14]- [18], [30], [36], [38]. These techniques include Convolutional Neural Network (CNN) [15], [16], [18], [36], Fully CNN (FCNN) [14], YOLO net v4 [17], Faster Region-based CNN (Faster R-CNN) [12], Generative Adversarial Network (GAN) [38], Mask R-CNN with Resnet [30], and Random Forests [18].…”

Sensor Fusion using Model Predictive Control for Differential Dual Wheeled Robot

“…A backbone layer and a detection layer make up YOLO. The output feature map is created by extracting features from the backbone layer [19]. Darknet is deployed as the backbone for the system.…”

Deep Transfer Learning Enabled Intelligent Object Detection for Crowd Density Analysis on Video Surveillance Systems

“…Moreover, the LR activation function is used for adding the non-linear factors to the CNN and exploiting dense blocks derived from crowd density estimation for calibrating the LR-CNN crowd density estimation model. Bhuiyan et al [22] developed a fully convolutional neural network (FCNN) model for crowd density estimation on surveillance video captured by a camera at a distance. Li et al [23] introduces a multi-scale feature fusion network (IA-MFFCN) depending upon reverse attention model that mapped the image into the crowd density map for counting purposes.…”