ScrapBoard

Ascend the scrap leaderboard

Comment

Inspiration

We wanted to make environmental sustainability more engaging and motivate people to actively participate in cleaning their communities. By combining trash collection with a competitive leaderboard, we created an app that encourages users to collect trash and see how they rank against others.

What it does

ScrapBoard is a mobile app where users can take pictures of trash, submit them for classification into one of six categories, and earn points. The app features a leaderboard where users can compete with friends, colleagues, and others in their community to see who can collect the most trash.

How we built it

We developed a mobile app for users to capture and submit trash images. These images are sent to a Flask server, where our machine learning model classifies them. To connect our servers and enable real-time testing, we used ngrok. We trained over 20 models to improve accuracy and prevent overfitting. The app also includes a leaderboard to gamify the experience.

Challenges we ran into

  • ngrok Setup: Configuring ngrok to establish smooth communication between the mobile app and the Flask server was a key hurdle.
  • Model Overfitting: We struggled with overfitting during training, forcing us to experiment with multiple models and techniques to improve generalization.
  • Frontend Design: Creating a clean and intuitive frontend while ensuring the app remained functional was a challenge.

Accomplishments that we're proud of

  • Successfully integrating ngrok for server communication across different environments.
  • Developing a machine learning model that accurately classifies trash into six categories after training over 20 different models, with differing learning rates and epochs.
  • Building a functional leaderboard system that encourages competition and drives engagement.

What we learned

We gained experience in server integration, machine learning, and mobile app development. The process taught us how to address issues like overfitting, optimize server communication using ngrok, and balance simplicity with functionality in the frontend.

What's next for ScrapBoard

We plan to expand ScrapBoard by adding features to track users' environmental impact, partner with local recycling centers, and improve the app’s model for even more accurate trash categorization. Additionally, we’re looking to scale the app to reach more users and inspire communities worldwide to take action against waste.

Built With

Share this project:

Updates