Recent Papers

Managing large crowds in public spaces such as railway platforms has become a critical challenge in ensuring public safety, smooth operations, and effective emergency response. In countries like India, where railway platforms witness extremely high passenger density on daily basis, the risk of overcrowding, panic induced movements, stampedes and security threats are ever present. Traditional surveillance system relies heavily on human operators who must monitor multiple CCTV feeds simultaneously, a task that is both exhausting and prone to delayed or inaccurate responses. To address these challenges, the Intelligent Crowd Management System For railway platforms Using Deep Learning introduces an automated and intelligent framework that leverages advanced computer vision and Deep Learning (DL) techniques to monitor, analyse, and respond to crowd dynamics in real time. The Proposed System integrates three major modules namely, Crowd Counting Module, Anomaly Detection Module, YOLO based Anomaly Localization Module. Firstly, The Anomaly Detection Module detects the anomaly through a convolutional autoencoder. Thirdly, The YOLO based Anomaly Localization Module pinpoints the exact location of anomalies within the video frame by drawing bounding boxes with confidence scores, offering railway authorities actionable insights. The Crowd Counting Module achieved a Mean Absolute Error (MAE) of 0.1513. The Anomaly Detection Module recorded a Mean Squared Error (MSE) of 0.1228. The YOLO based Anomaly Localization Module attained a mean Average Precision(mAP) of 0.5 of approximately 0.72 ensuring precise real-time detection and labelling of anomalous events through bounding boxes. The backend, built in Python with Flask, processes video streams and communicates with the Supabase database for storage, while the frontend dashboard is developed using ReactJS, Material UI and Recharts, provides a visually appealing interface for live monitoring, anomaly alerts, and historical data visualization. Experimental results show high accuracy across modules, confirming its practical effectiveness for enhancing safety and operational efficiency in crowded railway environments.

Copyright © 2025 Tamul Aiman Madrasi. This is an open access article distributed under the Creative Commons Attribution License.