AI Based Smart Bus Passenger Counting and Alert System for Bus Capacity

Abstract

Urban public transportation systems often face challenges related to overcrowding, passenger safety, and inefficient bus operations. This project proposes an AI- Based Smart Bus Passenger Counting and Alert System designed to monitor passenger load in real time. Using advanced computer vision techniques, the system detects and counts passengers entering and exiting the bus, maintaining an accurate record of occupancy. When the number of passengers approaches or exceeds the bus’s maximum capacity, the system generates instant alerts for the driver and central control, ensuring safety and compliance with regulations. Additionally, the system collects passenger data to assist transport authorities in route optimization and operational planning. By automating passenger monitoring, this solution enhances safety, improves service efficiency, and provides valuable insights for intelligent urban mobility management.

Introduction

Urban buses face chronic issues of overcrowding, passenger discomfort, and safety risks. Traditional passenger monitoring methods—manual counting, ticket collection, or simple sensors—are error-prone, labor-intensive, and fail to provide real-time data, making it difficult for transportation authorities to optimize routes, schedules, and fleet management.

Proposed Solution:
The AI-Based Smart Bus Passenger Counting and Alert System integrates computer vision, AI, and edge computing to monitor passengers entering and exiting in real time. Using cameras and deep learning models (e.g., YOLOv8) combined with tracking algorithms like Deep SORT, the system maintains an accurate, up-to-date count of passengers. When occupancy approaches or exceeds the bus’s maximum capacity, visual and audio alerts notify the driver and optionally a central control system. Historical data collection allows for analytics to optimize routes and schedules.

Objectives:

1. Real-time passenger detection and counting.

2. Continuous monitoring of bus capacity to prevent overcrowding.

3. Instant alerts for drivers and control centers when thresholds are exceeded.

4. Historical data collection for route optimization and demand prediction.

5. Reduction of manual monitoring and operational errors.

6. Scalable deployment across multiple buses integrated into smart city infrastructure.

Methodology & Implementation:

• Hardware: High-resolution RGB/RGB-D cameras at bus doors, edge computing devices (e.g., NVIDIA Jetson, Raspberry Pi) for on-board processing, and alert mechanisms for drivers.

• Data Acquisition: Continuous video capture and annotated datasets for AI model training.

• AI Model: Deep learning object detection (YOLOv8) for passenger identification and tracking (Deep SORT) to prevent double counting. Optimization includes background subtraction and multi-frame validation.

• Real-Time Processing: Edge device updates passenger count continuously as passengers enter/exit.

• Capacity Monitoring & Alerts: Passenger count compared against maximum capacity; alerts issued to driver and optionally to central control.

• Software & Integration: Python with OpenCV and PyTorch/TensorFlow for AI; bus interface displays real-time counts and alerts; system tested under diverse conditions for reliability.

Literature Insights:
Previous automated passenger counting systems used infrared, pressure mats, RFID, or simpler vision methods, but struggled with accuracy in crowded conditions and lacked integration with alert systems. AI and computer vision now allow precise real-time counting and capacity monitoring, addressing both operational and research gaps.

Impact:
This system improves passenger safety and comfort, reduces manual labor and human error, and supports data-driven, efficient urban bus management, making it a scalable solution for intelligent public transportation.

Conclusion

The AI-Based Smart Bus Passenger Counting and Alert System provides an efficient and reliable solution for monitoring bus occupancy in real time. By integrating advanced computer vision techniques and AI-based detection models, the system accurately counts passengers boarding and alighting, overcoming limitations of traditional methods such as infrared sensors or manual counting. The real-time processing capability ensures that bus drivers receive immediate alerts when the vehicle approaches or exceeds its maximum capacity, enhancing passenger safety and compliance with transportation regulations. Additionally, the system offers the potential for central monitoring, enabling fleet operators to optimize routes, prevent overcrowding, and improve overall operational efficiency. Overall, this project demonstrates the practical application of AI and edge computing in public transportation, addressing challenges like overcrowding, inaccurate passenger counts, and delayed response to capacity limits. The implementation of such intelligent systems contributes to smarter urban mobility, safer travel, and a more efficient public transit ecosystem. With further enhancements—such as integration with ticketing systems, predictive analytics for passenger flow, and multi-camera tracking—the system can evolve into a fully autonomous solution for modern smart cities.

References

[1] Prastyo A., Andzani S., Minhalina …, et al. Automatic Passenger Counting System on Public Buses Using CNN YOLOv8 Model for Passenger Capacity Optimization. International Journal of Information Engineering and Science (IJIES). [2] Nitti M., Pinna F., Pintor L., Pilloni V., Barabino B. iABACUS: A Wi-Fi-Based Automatic Bus Passenger Counting System. Energies, 13(6):1446, 2020. [3] Li F., Yang F-W., Liang H-W., Yang W-M. Automatic Passenger Counting System for Bus Based on RGB-D Video. Proceedings of the 2nd Annual International Conference on Electronics, Electrical Engineering and Information Science (EEEIS 2016). [4] Kim H., Sohn M-K., Lee S-H. Development of a Real-Time Automatic Passenger Counting System using Head Detection Based on Deep Learning. Journal of Information Processing Systems, Vol. 18, No. 3, pp. 428-442, 2022. [5] Wang P., Chen X., Chen J., Hua M., Pu Z. A Two-Stage Method for Bus Passenger Load Prediction Using Automatic Passenger Counting Data. IET Intelligent Transport Systems. [6] Labit-Bonis C., Thomas J., Lerasle F. Visual and Automatic Bus Passenger Counting Based on a Deep Tracking-by-Detection System. 2021. [7] Cherrier N., Rérolle B., Graive M., Dib A., Schmitt E. Context-Aware Automated Passenger Counting Data Denoising. arXiv preprint, 2024.

Copyright

Copyright © 2025 Sakshi Sambhaji Deshmukh, Nivedita Bhausaheb Shinde , Wagh Ankita Maruti, Prof. Rahane D. A.. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.