| Summary: | The rapid growth of urban populations and rising consumerism have led to an unprecedented increase in the volume and diversity of municipal waste, now exceeding 2 billion tons annually. A significant portion, at least one-third, is not managed in an environmentally safe manner, posing severe social and environmental challenges. Paper waste, constituting approximately 10% of municipal solid waste, is a notable component due to its high recyclability. Recycling paper can significantly conserve trees and reduce energy and water usage, providing substantial environmental benefits. This study aims to enhance waste management by utilizing image recognition and IoT technologies for the segregation of paper waste at the household level. Through the evaluation of two deep learning techniques, this research compare the performance of MobileNet and Restnet50 model for waste classification, assessed using accuracy, precision, recall, and F1-score metrics. Results indicate that the hybrid model combining ResNet50 and MobileNet outperforms standalone models, achieving an average accuracy of 97.90% on the first dataset and 73.49% on the more complex second dataset. These findings demonstrate the potential of integrating advanced machine learning and IoT technologies to improve the efficiency and sustainability of waste management practices. © 2024 IEEE.
|
|---|