BinBot

• 

• 

  Our Coding Process

• 

  Roboflow Training Our AI

• 

  We Used GitHub To Complete Our Project

Inspiration❗:

The idea for BinBot was inspired by the urgent need to tackle waste mismanagement and contamination in recycling systems. Every year, countless recyclable items are improperly sorted into landfills, while non-recyclable materials contaminate recycling streams, making the process inefficient and resource-intensive. Witnessing the environmental damage caused by this issue motivated us to develop a solution that combines education and smart technology to simplify waste sorting and encourage sustainable habits.

What It Does✍️:

BinBot is an AI-powered waste-sorting assistant designed to identify whether items are recyclable, compostable, or landfill-bound. Using a camera and a YOLOv9-based machine learning model trained with RoboFlow, BinBot analyzes objects through a webcam photo in real-time to classify them accurately. The system guides users with clear instructions on how to dispose of items through a user-friendly interface. Beyond sorting, BinBot provides educational feedback, such as tips on proper recycling practices and insights about specific items. By integrating real-time AI capabilities and practical education, BinBot makes waste management more efficient and accessible.

How We Built It🧑‍💻:

To build BinBot, we began by researching waste sorting challenges and identifying ways to address them using AI and computer vision. We used datasets from Kaggle to train our model, focusing on common waste items. Training the YOLOv9 model with Roboflow enabled it to achieve high accuracy in real-time object detection. The backend was developed in Python, leveraging Roboflow for image recognition, while the frontend used HTML, JavaScript and CSS to create a seamless interface. We integrated a webcam module for live scanning and used GitHub for version control and collaboration. To test BinBot, we conducted trials with diverse waste items, fine-tuning the model and interface based on feedback. The result is a lightweight, modular system that can easily retrofit onto existing bins, with a prototype mobile app to track personal recycling habits and provide sustainability tips.

Challenges We Ran Into:

Building BinBot was not without its challenges. Training the YOLOv9 model required extensive data preparation and fine-tuning to ensure it could handle ambiguous items like coated paper or mixed plastics. Integrating the AI model with the website and webcam posed additional difficulties, requiring significant debugging to achieve smooth real-time performance. Deploying the application involved ensuring the backend and AI models could run reliably on hosting platforms, which required careful setup and troubleshooting. Despite these challenges, our teamwork and dedication helped us overcome each hurdle and create a functional prototype.

Accomplishments That We're Proud Of🎆:

We’re proud of successfully training an AI model capable of accurately classifying waste items in real-time. Integrating the hardware and software into a fully functional prototype was another major achievement. The intuitive user interface we designed ensures that even non-technical users can benefit from BinBot. Additionally, our ability to collaborate effectively as a team, using GitHub to manage our workflow, stands out as a highlight of the project.

What We Learned👩‍🏫:

Through this project, we gained valuable experience in training machine learning models and understanding the complexities of real-time object detection with YOLOv9. We learned how to clean and organize datasets from Kaggle, integrate hardware and software components, and design user-centered interfaces. We were exposed to different codes in general, like how to make the webcam work. Most importantly, we deepened our understanding of sustainability and how technology can drive meaningful environmental change.

What's Next for BinBot🧑‍🔬:

Moving forward, we plan to expand BinBot's dataset to recognize more waste types, including regional variations in recycling rules. We aim to develop a more compact, durable, and cost-effective version of the hardware for broader use. The mobile app will be enhanced with features like waste reduction tips, upcycling ideas, and gamified eco-challenges to keep users engaged. Finally, we hope to partner with schools, offices, and public spaces to deploy BinBot on a larger scale, maximizing its environmental impact and helping communities embrace sustainable practices.

Built With

• css
• github
• html
• javascript
• kaggle
• python
• yolov9