Smart Pothole Detection and Mapping System using Deep Learning for NMC Application

Abstract

Potholes are a major cause of road accidents, vehicle damage, and traffic congestion in urban transportation networks. Traditional road inspection methods rely on manual surveys and citizen complaints, which are often inefficient and time-consuming. This paper proposes an intelligent pothole detection and mapping system designed for the Nagpur Municipal Corporation (NMC) using deep learning and Internet of Things (IoT) technologies. The proposed system integrates an edge computing device equipped with a camera module, GPS receiver, and GSM communication module to automatically detect potholes and transmit their geographical location to a central server. A YOLOv8 object detection model is employed to identify potholes from road images with high accuracy while maintaining real-time processing capability. Detected pothole locations are visualized on a Geographic Information System (GIS) dashboard that enables municipal authorities to monitor road conditions and prioritize repair operations. The proposed solution provides a scalable, cost-effective, and automated approach for road infrastructure monitoring, contributing to improved road safety and efficient urban maintenance management.

Introduction

The text describes an automated pothole detection and mapping system for urban road infrastructure, targeting the Nagpur Municipal Corporation (NMC). Rapid urban growth, heavy traffic, and aging roads cause potholes that damage vehicles, increase accident risks, and reduce transportation efficiency. Traditional manual inspections are labor-intensive, slow, and often incomplete.

Proposed System:

• Combines deep learning-based detection (YOLOv8) with GPS geolocation and a GIS visualization dashboard.
• Hardware: Installed on municipal vehicles (buses, garbage trucks) with a Raspberry Pi 4, 1080p USB camera, U-Blox NEO-6M GPS, and SIM800L GSM module for data transmission.
• Software Workflow:
  1. Data Collection & Model Training: Two-stage transfer learning—first on a large public dataset, then fine-tuned on local Nagpur road images.
  2. Real-Time Detection: YOLOv8 processes video frames continuously; potholes detected above a confidence threshold are geotagged.
  3. Data Transmission & Storage: Detection events are sent in JSON format to a cloud server and stored in a PostgreSQL/PostGIS database.
  4. GIS Dashboard: Web interface showing pothole locations, repair status, data filtering, and statistical analysis for maintenance planning.

Advantages:

• Continuous road monitoring without manual inspection
• Early pothole detection reduces accidents
• Data-driven maintenance improves efficiency and reduces labor costs
• Integration with smart city systems

Performance & Metrics:

• YOLOv8 achieves 20–25 FPS on Raspberry Pi 4 for real-time detection.
• Expected metrics: Precision, Recall, mAP (>90%), and system latency under 30 seconds.

Outcome:
A fully functional, scalable, and cost-effective system providing real-time pothole detection, geolocation, and actionable insights for municipal authorities, enabling faster road maintenance, improved safety, and predictive planning.

Conclusion

This paper presented a deep learning-based pothole detection and mapping system designed for smart city road maintenance. The proposed system integrates a YOLOv8 object detection model with an IoT-based hardware platform and a GIS visualization dashboard to provide an automated solution for road condition monitoring. By combining real-time image processing, GPS localization, and cloud-based data management, the system enables municipal authorities to efficiently identify and track pothole locations across the city. The proposed approach reduces manual inspection effort and improves the speed and accuracy of road maintenance operations. Future work will focus on enhancing the system by incorporating pothole severity estimation, drone-based road inspection, and predictive analytics for forecasting road damage. Integration with citizen reporting applications can also provide additional data sources to improve system accuracy and coverage.

References

[1] M. Sharma and P. Singh, \"Smart City Road Condition Monitoring using Accelerometer and GPS,\" IEEE Internet of Things Journal, vol. 9, no. 18, pp. 16945–16954, 2023. [2] R. Mishra, K. Jain and A. Patel, \"Design and Implementation of Low-Cost Pothole Detection System,\" in IEEE Sensors Conf., pp. 1–4, 2021. [3] S. Verma, A. Kumar and R. Gupta, \"Pothole Detection and Warning System using IoT Sensors,\" IEEE Access, vol. 10, pp. 105231–105242, 2022. [4] J. Dharneeshkar et al., \"Deep Learning based Detection of Potholes in Indian Roads using YOLO,\" International Conference on Inventive Computation Technologies (ICICT), 2020. [5] E. N. Ukhwah, E. M. Yuniarno and Y. K. Suprapto, \"Asphalt Pavement Pothole Detection using Deep Learning based on YOLO Neural Network,\" ISITIA, 2019. [6] H.-W. Wang et al., \"A Real-Time Pothole Detection Approach for Intelligent Transportation Systems,\" Mathematical Problems in Engineering, 2015. [7] A. Johnson and B. Gupta, \"A Comparative Study of Object Detection Algorithms for Road Hazard Detection,\" Journal of Intelligent Transportation Systems, vol. 25, no. 4, pp. 415-430, 2021. [8] F. Ahmed, \"3D Pothole Reconstruction using Stereoscopic Vision for Autonomous Vehicles,\" IEEE Transactions on Intelligent Vehicles, vol. 6, no. 2, pp. 257-267, 2021. [9] P. Singh and R. Kumar, \"An IoT-Based Framework for Smart City Road Maintenance Management,\" in Proc. International Conference on Smart Systems and Inventive Technology (ICSSIT), pp. 112-117, 2022.

Copyright

Copyright © 2026 Dr. Ranjit Bobate, Kalash Gajghate, Rohit Deshmukh, Yogendra Meshram, Mayur Nande. 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.