Comparative Analysis of Deep Learning Models for Ai-Driven Smart Waste Classification System Using Resnet, Efficientnet, and VGG16 for Automated Waste Segregation

Abstract

Effective waste management is critical for environmental sustainability and public health. Traditional waste segregation methods rely heavily on manual sorting, which is time-consuming, error-prone, and hazardous for workers. This paper presents a comprehensive comparative analysis of three state-of-the-art deep learning architectures—ResNet-50, EfficientNet-B0, and VGG16—for automated waste classification. The models are trained to categorize waste into six primary classes: Cardboard, Glass, Metal, Paper, Plastic, and Trash. Our experimental evaluation demonstrates that EfficientNet-B0 achieves the highest performance with a test accuracy of 96.8%, followed closely by ResNet-50 at 96.6% and VGG16 at 93.1%. EfficientNet-B0 also demonstrates superior training efficiency, reaching 95% accuracy in just 22 epochs compared to 25 epochs for ResNet-50 and 35 epochs for VGG16. The F1-scores across all waste categories range from 0.93 to 1.00 for EfficientNet-B0, indicating robust classification performance. This comparative study provides valuable insights for selecting appropriate deep learning architectures for real-world waste management applications in smart cities and recycling facilities.