DEVELOPMENT OF A MACHINE LEARNING-BASED WASTE CLASSIFICATION SYSTEM USING VGG-16 CNN FOR ENHANCED BIODEGRADABLE AND NON-BIODEGRADABLE SEGREGATION

Abstract

Waste management struggles with the segregation of biodegradable and non-biodegradable waste due to improper disposal, leading to contamination and reduced recyclable material quality. This study addresses these challenges by developing a machine learning-based waste classification system. Utilizing image classification techniques, specifically the VGG-16 Convolutional Neural Network (CNN) model, the system categorizes waste into biodegradable and non-biodegradable using a dataset from an open-source website. The methodology includes eight phases: preliminary study, knowledge and data acquisition, data pre-processing, model design, development, testing, and evaluation. A prototype using Maker Uno and a servo motor physically demonstrates waste classification. Despite challenges like limited high-quality components, this study aims to enhance recycling efficiency and sustainability by using VGG-16 with different epochs and shows the prototype's effective functionality, offering a promising solution for improving waste segregation and management.