<![CDATA[The rapid advancement of Deep Learning has significantly contributed to the development of Smart Waste Classification systems for improving automated waste sorting and sustainable waste management. However, comprehensive studies that systematically summarize the evolution of Deep Learning models, datasets, evaluation methods, and future research directions remain limited. This study aims to conduct a Systematic Literature Review (SLR) on Deep Learning models applied to Smart Waste Classification by following the PRISMA 2020 guidelines. The literature search was performed using Publish or Perish across several scientific databases, covering publications from 2020 to 2025. A total of 100 studies were initially identified, and after the identification, screening, eligibility, and inclusion processes, 40 studies were selected for qualitative synthesis. The findings indicate that Convolutional Neural Network (CNN) remains the most widely adopted architecture, followed by ResNet, MobileNet, EfficientNet, YOLO, and Vision Transformer. TrashNet is the most frequently used dataset, while accuracy, precision, recall, F1-score, and mean Average Precision (mAP) are the dominant evaluation metrics. Current research trends emphasize transfer learning, lightweight architectures, and the integration of Deep Learning with the Internet of Things (IoT) and edge computing. This review provides comprehensive insights into recent developments and identifies research opportunities for developing more accurate, efficient, and practical Smart Waste Classification systems. Keywords— Systematic Literature Review (SLR); Deep Learning; Smart Waste Classification; Convolutional Neural Network (CNN); Computer Vision; Transfer Learning; Waste Management.]]>
Copyrights © 2026
