Metode Systematic Literature Review Arsitektur Convolutional Neural Network untuk Klasifikasi Sampah Berbasis Citra

Abstract

Waste management is becoming increasingly complex as the volume and variety of waste types increase, making manual sorting processes inefficient and error-prone. In recent years, image-based waste classification using Deep Learning (DL), particularly Convolutional Neural Networks (CNN), has been widely developed as an automated waste sorting solution. This study aims to conduct a Systematic Literature Review (SLR) to analyze the performance, trends, and utilization of CNN architectures in image-based waste classification. The SLR method is carried out systematically through the stages of planning, literature search, study selection, data extraction, and result synthesis. A total of 15 scientific articles published between 2020 and 2025 were analyzed with a focus on CNN architecture, preprocessing techniques, dataset characteristics, number of classes, and evaluation metrics such as accuracy, precision, recall, and F1-score. The results of the study show that CNNs are capable of providing high classification performance in various scenarios. Standard CNNs are effective on small datasets with a limited number of classes, while transfer learning approaches such as MobileNetV2, ResNet50, VGG16, and EfficientNet show more stable performance in multi-class classification. Overall, CNN proved to be an effective and adaptive approach to support computer vision-based automated waste sorting systems.