Abstract :

Road Infrastructure Deterioration, Particularly Pothole Formation, Poses Critical Threats To Public Safety And Economic Productivity In India, With Potholes Alone Accounting For Over 3,600 Accidents And 1,500 Deaths In 2021. Existing Monitoring Approaches—manual Periodic Inspections And Vehicle-based Survey Systems—suffer From Temporal Coverage Gaps, Spatial Blind Spots, Subjective Variability, And Absence Of Real-time Alerting, Leaving Road Damage Undetected For Days Or Weeks After Occurrence. This Paper Presents An Automated Road Damage Detection System That Transforms Existing Municipal CCTV Infrastructure Into A Proactive, Continuous Monitoring Network Using A Custom-trained YOLOv8 Deep Learning Model. The Proposed System Connects To Multiple Simultaneous IP Camera Streams Via RTSP/HTTP Protocols Using OpenCV VideoCapture, Processes Frames Through A YOLOv8 Model Trained On 6,000 Labeled Road Damage Images Achieving 96.4% MAP50, 96.8% Precision, And 95.2% Recall Across Four Damage Classes (pothole, Longitudinal Crack, Transverse Crack, Alligator Crack). A Thread-safe Multi-camera Management Module With Exponential-backoff Reconnection Ensures 24/7 Uninterrupted Operation. Upon Detecting Damage With Confidence Exceeding 96%, The System Automatically Dispatches HTML-formatted SMTP Email Alerts With Detection Images, Timestamps, And Confidence Scores To Designated Municipal Authorities, Closing The Detection-tonotification Loop. A Python Tkinter Desktop Application Provides Real-time Visualization Of Camera Feeds With Bounding Box Overlays, Detection Logging To SQLite, And System Statistics. Comprehensive Evaluation Comprising 35 Unit Tests, Integration Tests, System Tests, And User Acceptance Testing With Municipal Professionals Yields 98.1% Overall Test Pass Rate And 4.7/5 User Satisfaction. Frame Processing Averages 156ms Per Inference, Enabling Effective Real-time Monitoring Of 4–6 Simultaneous Camera Streams On Recommended Hardware, Representing A Cost-effective Solution Requiring No New Camera Infrastructure Investment.

Published:

02-4-2026

Issue:

Vol. 26 No. 4 (2026)

Page Nos:

202-208

Section:

Articles

License:

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

How to Cite