GreenLens: An AI-Powered Waste Classification Application

Abstract

Have you ever finished a meal but were unsure which recycling container to throw away your trash in? You might not know where to throw away your napkins, plates, utensils, leftovers, etc. Not recycling correctly can have devastating environmental damage. Plastic pollution from neglecting to recycle is one of the main causes of ocean pollution, destroying much marine life. This AI-powered waste classification app called GreenLens is able to recognize dozens of common daily wastes that people have little knowledge of how to dispose of properly. GreenLens, using people’s phone cameras, is easily accessible to the majority, has the ability to scan and recognize the type of waste in a matter of seconds, and gives detailed instructions on how to dispose of the waste. It is lightweight, yet powerful, allowing it to achieve recognition accuracies of up to 95%. With the help of GreenLens, people will be able to identify common daily wastes easily and dispose of them correctly, thus helping to reduce pollution caused by inappropriate recycling. In this paper, I will talk about how I came up with the idea of developing this application, as well as the technologies that were applied to build GreenLens. In many experiments, GreenLens is able to recognize all wastes trained for the model while also achieving high accuracies.