EfficientNet-Based Deep Learning Model for Advanced Waste Classification

Paper ID: 39

Abstract

This study focuses on developing an autonomous waste sorting system for Bangladesh by using intelligent image classification for waste categorization. A dataset of 5,012 waste images was classified into nine categories, and pre-trained EfficientNet models (B0, B2, B3, and B4) were used for feature extraction. The models achieved impressive results, with EfficientNet B0 and B2 reaching 93% accuracy, and EfficientNet B3 and B4 achieving 91%, outperforming conventional classification methods.

Keywords