Inspiration

Road damage such as potholes and cracks is a common problem that affects road safety and transportation. Traditional road inspection is often time-consuming and depends on manual observation. We wanted to create a smarter solution using Artificial Intelligence to make road inspection faster, easier, and more organized.

What it does

Road Guardian AI is an AI-powered road damage detection system that uses computer vision to identify visible road defects from images and videos.

The system aims to:

Detect potholes and different types of road cracks.
Identify and classify visible road damage using a YOLO-based model.
Display detected defects with bounding boxes and confidence scores.
Estimate damage severity using rule-based logic.
Generate structured inspection reports.
Help prioritize road maintenance based on detected damage.

How we built it

We designed the project using the following technologies:

Python: Main programming language.
YOLO: Object detection for identifying road damage.
OpenCV: Image and video processing.
Streamlit: Interactive user interface.
SQLite: Storing inspection records.
Pandas: Data analysis and management.
Folium: Map-based visualization with optional GPS information.

Our planned workflow is to capture or upload road images, process them using OpenCV, detect defects using YOLO, assess their severity, and display the results through a dashboard.

Challenges we ran into

Some of the major challenges include identifying different types of road damage, handling variations in lighting and image quality, selecting suitable training data, and determining consistent severity levels.

We address these challenges through image preprocessing, suitable model training, and transparent severity classification rules.

Accomplishments that we're proud of

Identified a real-world problem affecting road safety and transportation.
Developed an AI-powered solution concept for detecting potholes and road cracks using computer vision.
Designed an end-to-end workflow combining image processing, YOLO-based detection, and damage severity assessment.
Planned a user-friendly dashboard to display road damage detection results and inspection reports.
Explored how AI can support smarter road maintenance and help prioritize repairs.

Worked collaboratively as Team ASTRA, combining our ideas and technical knowledge to develop a solution with real-world potential.

We are proud that our project brings together AI and practical problem-solving to contribute towards safer roads and smarter infrastructure.

What we learned

Through this project, we explored computer vision, object detection, image processing, AI model development, data management, and dashboard development. We also learned how AI can be applied to real-world problems to support better decision-making.

What's next for Road Guardian AI

Our future plans include:

Improved AI Accuracy: Train and improve the YOLO model using more diverse road damage images.
GPS Integration: Automatically record the location of detected road damage.
Live Camera Detection: Enable real-time road damage detection using a camera.
Mobile Application: Develop a mobile-friendly version for easier road inspections.
Smart Maintenance Priority: Improve severity assessment to help authorities identify roads that need urgent repairs.

6.Historical Monitoring: Compare repeated inspections to track how road conditions change over time.