Enhancing Plastic Recycling through Machine Learning and Computer Vision

A Case Study on Plastic Bottles

Abstract

Enhancing recycling efficiency remains a challenge due to many contemporary innovations’ cost and organizational constraints. Conventional methods of plastic categorization are not scalable for plastic recycling due to extreme costs for multiple sets of hardware, such as different types of sensors. However, in this project, I propose a new method of classifying plastic based on computer vision. As a case study, I decided to focus on creating a model for identifying different kinds of plastic in plastic bottles. I used this model to increase accuracy of identifying and classifying different kinds of bottles, such as clear and blue plastic bottles, and was able to find that with a neural network, the model could classify the bottle at a high rate compared to the baseline. The results of my study show the potential of this technology in context of economically sustainable recycling. My model displays an ability to distinguish between different types of bottles, surpassing my baseline by a significant margin of 58.83%. This improvement in accuracy has large implications for improving recycling efficiency and plastic waste management. These findings highlight the potential of integrating machine learning with computer vision for cost-effective and efficient recycling methods.