Machine Learning for Plastic Waste Detection: State-of-the-art, Challenges, and Solutions
Owen Tamin, Ervin Gubin Moung, Jamal Ahmad Dargham, Farashazillah Yahya, Sigeru Omatu, Lorita Angeline
Abstract
Machine learning has advanced rapidly in recent technologies of artificial intelligence. The abundant and cheap computation in machine learning has created more knowledge and reliability of this method. Therefore, more researchers have relied on machine learning to solve plastic waste pollution, which contributes to global greenhouse gas emissions. Datasets are fundamental requirements to build the foundation of the plastic waste detection system. This paper provides an overview of several state-of-the-art deep learning models, focusing on the dataset used for waste classification and identification. The performance of each deep learning model in classification accuracy is also discussed. The challenges of implementing deep learning in plastic waste detection are identified; (i) Data deficiency for training, (ii) Quality of the existing plastic waste dataset, and (iii) Color spaces. Lastly, the proposed solution to each challenge is presented.