Environmental impacts in e-waste management using deep learning
Godfrey Perfectson Oise, Susan Konyeha
Abstract
This paper introduces a robust neural network framework to address the environmental difficulties of e-waste management by combining EfficientNet, MobileNet, and a Sequential Neural Network (SNN) for effective classification and sorting of different e-waste categories. The suggested model showed an impressive 98% accuracy, with precision, recall, and F1-scores of 98%, 97%, and 97%, respectively, surpassing standalone models such as MobileNet (86%), EfficientNet (85%), and Yolov8 (88%). ROC analysis demonstrated exceptional performance, with an AUC of 1.00 for all classes. Advanced preprocessing techniques, including SMOTE and data augmentation, ensured dataset balance and improved model generalization. The framework addresses challenges like feature overlap between classes and demonstrates scalability and adaptability for real-world applications. This innovative approach enhances recycling efficiency and promotes environmental sustainability, providing a pathway for deploying AI-driven solutions in e-waste management.