A Proposed Model for Crowd Management Using Deep Learning in Al-Masjid Al-Nabawi

Abstract

The rituals of Hajj and Umrah attract millions of people to Makkah and Madinah annually, making them the largest human gathering in the world, with nearly three million participants. Managing the safety of such massive crowds is a critical issue, as the risks of stampedes and overcrowding pose significant threats to pilgrims and visitors. With Saudi Arabia striving to become a global leader in the tourism sector, the importance of effective crowd management has become greater than ever to ensure a safe and seamless experience for millions of visitors. The proposed model consists of three main phases. First, video and image data are collected and subjected to pre-processing to improve their quality and ensure suitability for deep learning analysis. Second, a deep neural network (DNN) is trained to analyze crowd movement, density, and potential risks. Third, based on classifying crowd density into different levels (for example, from Normal to Heavily Crowded), appropriate management actions can be activated. For instance, if a particular courtyard is identified as having high density, the system can suggest redirecting worshippers to alternative gates or trigger an alert to crowd control teams. Conversely, a low-density classification indicates that conditions are normal. Through this analysis and automated response, the system aims to enhance the safety and comfort of worshippers, while providing strong support to mosque authorities, especially during peak crowding periods.