• 

  Pothole :[

INSPIRATION

Our inspiration stemmed from the persisting problem of potholes and the frustration of waiting for government intervention. Cities are riddled with horrible road conditions, and drivers are left at the mercy of city officials to find, report and fix potholes. We aimed to address this issue with innovation, utilizing AI to empower users to streamline the first step of the process: reporting the issues.

WHAT IT DOES

PaveGuard is a user-friendly mobile app that revolutionizes pothole reporting. Users can instantly scan potholes which are then uploaded to a database. Our trained model then classifies the images by obstruction type (pothole, fallen tree, crack, etc) and then uploads the images and the user provided data and geodata to an admin console. City officials can then log in and access these images, where they can then take action on these potholes in a much more proficient manner. This reduces the $3 billion spent annually by drivers on pothole-related damages, which could be much better used in America's growing cities to address social issues.

HOW WE BUILT IT

We employed a robust tech stack for PaveGuard. The front end was developed using React and Bootstrap, ensuring a visually appealing and intuitive user interface. Our backend relied on Firebase and Python, enabling seamless integration with our custom-trained AI, based on YOLO (You Only Look Once). The AI model was trained using a dataset from Kaggle, enriched with data augmentation techniques like image stretching and rotations.

CHALLENGES WE RAN INTO

Our primary challenge was dealing with diverse image sizes when passing them to the AI model, which was configured for 619x619 images. This resulted in issues like image distortion and inaccurate predictions. We overcame this through model adjustments and preprocessing. Additionally, the tight time frame and AI model training posed difficulties. Another challenge was fetching data between the front end and backend, which we overcame by switching from MongoDB to Firebase.

ACCOMPLISHMENTS WE'RE PROUD OF

We're proud of creating a well-designed, accessible app that simplifies pothole reporting. Our app's ease of use, real-time image classification, and connection to authorities showcase our commitment to solving a real-world problem effectively.

WHAT WE LEARNED

This hackathon taught us the importance of effective collaboration within a team. We also gained insights into optimizing AI models for real-world applications, addressing image size variations, and the challenges of real-time processing.

WHAT'S NEXT FOR PAVEGUARD

In the future, we aim to collaborate with local and state governments, incentivizing users to report road issues by offering monetary rewards. We also plan to expand our app's functionality, addressing a broader range of road-related problems beyond just potholes. We want to expand to better accessibility for mobile users as well.

Built With

• bootstrap
• css3
• firebase
• html5
• javascript
• python
• react