Garbage Vision is a very simple iOS app that I wrote for identifying trash. Given an image, Garbage Vision classifies the subject into one of six possible classes: Cardboard, Glass, Metal, Paper, Plastic, and Trash. The app is written in Swift and its UI is implemented using the Swift UI framework. Classification is handled... poorly... by the Garbage Net Core ML model, which I created using Apple's Create ML app and trained using a trash image dataset from the trashnet GitHub repo by Gary Thung. In practice, classification seemed to be hit or miss, especially in differentiating between Paper and Plastic.
I was inspired to create Garbage Vision after seeing the various compost, recycle, and trash signs at my workplace. The idea of a mobile app that could visually identify trash sounded intriguing as a logical extension. Also, I had nothing better to do on Christmas due to the ongoing COVID-19 pandemic >:(
