ARCHIVES

Passenger congestion in railway coaches is becoming increasingly frequent due to rising urban populations and heavy reliance on rail transportation, especially during peak hours. High crowd density affects passenger comfort and also creates safety risks such as limited mobility, delayed emergency evacuation, and operational inefficiencies. Modern monitoring systems often rely only on camera-based counting or manual supervision. However, these approaches are sensitive to lighting changes, visual blockage, and network delays, making them less reliable in real-time coach environments. Another major challenge is maintaining consistent occupancy assessment when passenger movement, occlusion, and environmental variations occur simultaneously. Traditional single-sensor solutions are too limited to provide stable and trustworthy crowd condition evaluation. This paper presents an onboard crowd and passenger occupancy detection system that integrates visual sensing with multiple non-visual physical sensors and environmental monitoring. The system uses embedded processing and multi-sensor data fusion to analyze passenger presence and determine qualitative crowd states in real time. By combining camera analysis with hardware sensor validation, the system improves detection robustness under dynamic and low-visibility conditions. All processing and decision making are performed locally on the embedded unit, enabling continuous operation without dependence on remote servers. The system also generates immediate visual and audio alerts inside the coach to support operational awareness and safety response.