BinSense

Inspiration

In our school I always noticed two cans next to each other, the trash and recycling. Even though the recycling can is right there I would always see paper, bottles, and other things that should belong in the recycling can ending up in the trash.

What it does

This project uses a computer running YOLOv5(object detection model) to automatically check whether an object is trash or recycling then sends the signal to the arduino where it uses a servo to open up the trashcan

How we built it

(Dataset: https://universe.roboflow.com/grpglmthesis/thesis-unztx/dataset/19)

Computer part: Yolov5 + OpenCV Pyserial

Hardware: Cardboard and painters tape Arduino Uno 2 x LED 2 x Servos

Challenges we ran into

Some challenges we ran into was that the

• Model had a lot of false accuracy so I had to tone down the confidence percentage.
• The stepper motors were very weak so instead of a cardboard top I had to use paper

Accomplishments that we're proud of

Being able communicate using Arduino and yolov5

What we learned

I learned how to tune models even after training so that it could accurately find

What's next for BinSense

• Switching to a raspberry pi so that everything can run directly on site
• Making our own accurate dataset

Built With

• arduino
• pyserial
• yolov5
• zed