Crowd Density Estimation in Railway Coaches

Abstract

Crowd Density Estimation in Railway Coaches with the growing demand for public transportation and with the increasing use of unreserved coaches, overcrowding has become a major concern regarding passenger safety, comfort, and operational control. Traditional methods of monitoring crowd levels rely heavily on manual observation, which is ineffective and leads to errors. Currently, no effective crowd monitoring system is used in these compartments. The program processes captured images using Histogram of Oriented Gradients (HOG) to extract important visual patterns. The extracted features are classified into low, medium, and high crowd levels using multiple machine learning models trained and tested on railway coach images collected under different crowd density conditions. Performance is evaluated using Accuracy, Precision, Recall, and F1-score, with the MLP Neural Network achieving the highest accuracy of 94.7%. Whenever high crowd density is detected, an email alert along with the captured image is automatically sent to the railway authorities for timely action. By combining HOG-based feature extraction, multiple machine learning models, and an automated email alert mechanism, the proposed system provides a practical, cost-effective, and scalable solution for real-time crowd monitoring, improving passenger safety and efficient crowd management inside railway coaches.