Waste management is a complex and dynamic issue that demands creative and visionary solutions that exploit the potential offered by recent technological advances. Our research investigates the application of machine learning and deep learning in image recognition and categorization within the waste spectrum. We trained a Convolutional Neural Network (CNN) model on a massive dataset of pictures depicting organics, among others, typically generated as recyclables. We aim to develop a classification model for organic and recyclable waste that leverages transfer learning to classify them with high accuracy. This study aims to lay the foundation for future systems that will recycle automatically while improving waste recycling processes, hence reducing environmental impacts. The goal of our research was to create an image classification model that could differentiate photos between organic and recyclable waste by designing a classifier using the VGG16 architecture. Our study utilized the VGG16 model, based on Convolutional Neural Networks (CNNs), to achieve a precision score of 0.96 for organic and 0.88 for recyclable. This indicates that our model effectively reduces incorrect predictions for non-categorized items. The model achieved a high recall rate of 0.97 on pictures of recyclable garbage, indicating that it could identify most "Recyclable" examples properly. Moreover, these results highlight the VGG16 architecture's effectiveness in categorizing trash types, indicating potential room for improvement in recognizing "Organic" garbage images by the model, particularly in terms of recall.
Shafek et al. (Sun,) studied this question.