recycle buddy

Inspiration

The growing need for efficient recycling methods inspired us to develop an automated system that could identify and sort recyclables. We were particularly motivated by Google's X moonshot projects, such as their efforts to tackle the plastic crisis through advanced technologies x.company . These initiatives demonstrated the potential of combining machine learning with practical applications to address environmental challenges.

Learning

Through this project, we deepened our understanding of object detection algorithms, specifically YOLO (You Only Look Once). We learned how to train and implement YOLO models for real-time waste detection, enhancing our skills in machine learning and computer vision. Additionally, we gained experience in integrating various microcontrollers and machine learning models to create a cohesive system.

Project Development

We began by collecting a dataset of recyclable items and training a YOLO model to recognize them. The trained model was then integrated with microcontrollers to control sorting mechanisms. We designed a physical mock-up of the sorting system using duct tape and other materials to demonstrate the concept. This prototype allowed us to test and refine the system's functionality in a controlled environment.

We manually trained a YOLOv8 model in Google Colab with hand-annotated images.

Challenges

One of the main challenges was creating a functional prototype with limited resources. Constructing the physical mock-up out of duct tape required creativity and problem-solving to ensure it could effectively demonstrate the sorting process. Another challenge was interfacing the various microcontrollers with the machine learning models. Ensuring seamless communication between hardware and software components required careful planning and troubleshooting.

Built With

• esp32
• motor
• orangepi
• yolov8