Khakam Ma'Ruf, Apry Aditya Saputra, Rizal Justian Setiawan, Nur Azizah, Darmono
The growing volume of solid waste in urban areas has increased the demand for efficient and reliable waste management systems, particularly in the waste sorting stage, which is still largely dependent on manual methods. Manual sorting is time-consuming, costly, and prone to human error, thereby reducing recycling efficiency and increasing environmental risks. To address these challenges, this study proposes an automated waste classification approach based on Deep Learning using a Convolutional Neural Network (CNN) architecture. The proposed model is designed to classify waste images into six categories, namely cardboard, glass, metal, paper, plastic, and trash, using a dataset consisting of \mathbf{2, 5 2 7} images. To mitigate the effects of class imbalance and enhance model generalisation, data augmentation techniques are applied during the training process. The CNN architecture comprises three convolutional blocks integrated with MaxPooling and Dropout layers to optimise feature extraction and reduce overfitting. Model training is further enhanced through the use of the Adam optimiser, Early Stopping, and ReduceLROnPlateau strategies to improve convergence stability and learning efficiency. Experimental results indicate that the proposed model achieves a best validation accuracy of 66.31%. Although performance on minority classes remains limited due to dataset imbalance, the model demonstrates a strong ability to learn discriminative features and accurately classify dominant waste categories. The findings of this study highlight the potential of CNN-based automated waste classification as a foundational component of smart waste management systems, with prospective applications in automated sorting facilities, intelligent recycling systems, and sustainable environmental management initiatives. © 2026 IEEE.
Universitas Gadjah Mada, Faculty of Engineering, Dept of Industrial Engineering, Yogyakarta, Indonesia; Universitas Gadjah Mada, Faculty of Engineering, Dept of Information Technology, Yogyakarta, Indonesia; National Chung Hsing University, Faculty of Law and Politics, Dept of Asia and China Studies, Taichung, Taiwan; School of Medicine, National Cheng Kung University, Dept of Public Health, Taian, China; Universitas Negeri Yogyakarta, Faculty of Engineering, Dept of Civil Engineering Education, Yogyakarta, Indonesia
