Deep Dilated Convolutional Neural Network for Crowd Density Image Classification with Dataset Augmentation for Hajj Pilgrimage

Almost two million Muslim pilgrims from all around the globe visit Mecca each year to conduct Hajj. Each year, the number of pilgrims grows, creating worries about how to handle such large crowds and avoid unpleasant accidents or crowd congestion catastrophes. In this paper, we introduced deep Hajj crowd dilated convolutional neural network (DHCDCNNet) for crowd density analysis. This research also presents augmentation technique to create additional dataset based on the hajj pilgrimage scenario. We utilized a single framework to extract both high-level and low-level features. For creating additional dataset we divide the process of images augmentation into two routes. In the first route, we utilized magnitude extraction followed by the polar magnitude. In the second route, we performed morphological operation followed by transforming the image into skeleton. This paper presented a solution to the challenge of measuring crowd density using a surveillance camera pointed at a distance. An FCNN-based technique for crowd analysis is included in the proposed methodology, particularly for classifying crowd density. There are several obstacles in video analysis when there are a large number of pilgrims moving around the tawaf area, with densities of between 7 and 8 per square meter. The proposed DHCDCNNet method has achieved accuracy of 97%, 89% and 100% for the JHU-CROWD dataset, the UCSD dataset and the proposed Hajj-Crowd dataset, respectively. The proposed Hajj-Crowd dataset, the UCSD dataset, and the JHU-CROW dataset all had accuracy of 98%, 97% and 97%, respectively, using the VGGNet approach. Using the ResNet50 approach, the proposed Hajj-Crowd dataset, the UCSD dataset, and the JHU-CROW dataset all had an accuracy of 99%, 91% and 97%, respectively.