Transfer Learning Application for an Electronic Waste Image Classification System

E-waste or WEEE poses a growing environmental problem owing to the presence of toxic substances. The improper disposal of e-waste can damage the environment and human health. E-waste management involves proper collection, transportation, storage, processing, and disposal of waste. Effective e-waste management requires an accurate classification to separate recyclable items from hazardous materials. Manual categorization can be time-consuming, but automated methods such as artificial learning offer promise. This chapter explores the use of artificial learning algorithms, including machine and deep learning, for e-waste classification. Various techniques such as decision trees, random forest classifiers, support vector machines, and convolutional neural networks (CNN) are employed depending on the dataset and e-waste characteristics. CNN architectures that leverage transfer learning from large datasets, such as ImageNet, have shown promising results. This chapter presents a framework for modeling and classifying e-waste and compares the performances of different CNN architectures using metrics such as accuracy, recall, precision, and f1. A dataset of e-waste images from ten categories was collected, and transfer learning was applied to the selected architectures. The proposed intelligent classification system contributes to sustainability by automating e-waste categorization and reducing time and labor.