Hajj & Umrah Crowd Management Using Artificial Intelligence & Augmented Reality

Abstract The annual Hajj and Umrah pilgrimages attract millions of pilgrims, posing significant challenges in crowd management and safety assurance. This paper introduces an Artificial Intelligence (AI) and Augmented Reality (AR)-based system leveraging a structured workflow for efficient crowd management. Using a CSV dataset containing detailed crowd movement and behavior information, the system applies robust data pre-processing methods, including handling missing values with interpolation, detecting outliers via z-scores, and scaling data with standard scalar to ensure consistency. Feature extraction employs Principal Component Analysis (PCA), a dimensionality reduction technique that transforms high-dimensional data into a lower-dimensional space while preserving maximum variance. Feature selection is performed using Recursive Feature Elimination (RFE) and mutual information-based methods to optimize performance. Classification is performed using a Stacked Sequential Gradient Decision Boosted Convo Net (SSGDB-CN) Classifier, which enhances adaptive learning by sequentially refining predictions through gradient-based boosting while leveraging deep spatial feature representations. AR further enhances the system by providing dynamic navigation and live guidance to pilgrims, reducing congestion and improving experiences. Implemented using Python libraries such as Pandas, Scikit-learn, and TensorFlow, the proposed system demonstrates its potential to revolutionize traditional crowd management with advanced AI (accuracy – 98%, precision, recall, F1-score, and AUC) and AR technologies to determine dynamic navigation, guidance to pilgrims, reducing congestion, and improving overall experiences.