AUTOMATIC ROAD EXTRACTION AND ALERT GENERATION

Project Code :TCMAPY2114

Objective

The primary objective of this project is to develop an automated system for detecting and classifying road damages using deep learning. It aims to accurately identify multiple types of damages, including potholes, cracks, rutting, and patching. Another objective is to perform road extraction from images to focus analysis on relevant regions and improve detection efficiency. The system seeks to provide static maintenance recommendations based on damage type and severity. It also aims to implement an email-based alert mechanism to notify authorities of critical road conditions. The project intends to offer a user-friendly interface through Streamlit for visualization and monitoring. Enhancing inspection speed and reducing human error are key goals. Ultimately, the project strives to support proactive road maintenance and improve transportation safety.

Abstract

Road infrastructure monitoring is crucial for ensuring traffic safety and minimizing maintenance costs. This project proposes a deep learning–based system for automated road extraction and damage detection using YOLOv9. The model is trained on the Roboflow Road Damage Detection dataset, which includes seven types of road damages: alligator cracking, edge cracking, longitudinal cracking, transverse cracking, patching, potholes, and rutting. High-resolution road images are analyzed to accurately localize and classify surface defects in real time, while road extraction isolates road regions for focused analysis. Detected damages are further assessed to generate static maintenance recommendations based on severity and type. The system significantly improves inspection efficiency compared to manual surveys. A Streamlit-based front-end provides an interactive and user-friendly visualization of detection results, while the Python back-end runs on Google Colab for scalable training and inference. Additionally, an automated alert mechanism sends email notifications to the administrator whenever critical damages are detected, ensuring timely intervention and proactive maintenance. The proposed solution enhances accuracy, reduces operational costs, and supports the development of smart and sustainable road infrastructure.

Keywords: Road Damage Detection, YOLOv9, Deep Learning, Road Extraction, Pothole Detection, Crack Classification, Alert Notification System, Streamlit, Python, Smart Infrastructure.

NOTE: Without the concern of our team, please don't submit to the college. This Abstract varies based on student requirements.