Application of Deep Convolutional Neural Networks (DCNNs) for Automated Classification of E-Waste Components: Focusing on Circuit Boards, Batteries, and Cables

The proliferation of electronic waste (e-waste) has become a pressing environmental concern, necessitating efficient methods for its classification and management. Though there are traditional methods which are used when sorting and recyling e-waste, they are time-consuming, costly and most of the time, are not accurate and hence, a lot of resources are wasted in the process. This research aims at extending the use of Deep Convolutional Neural Networks (DCNNs) in the classification of critical e-waste components namely circuit boards, batteries, and cables. The proposed method fully utilizes the capacity of extracting features of DCNNs in order to accurately detect and sort these components by means of the receiving visual information. For training and testing the DCNN model a dataset was created and preprocessed and the target images included circuit boards of different orientations and sizes, batteries and cables. The architecture of a network was created considering aspect of depth and computational complexity guarantying high accuracy and feasible time for great data set processing. In the evaluation process of the model, a comparison was made with various standard performance measures, with checklists such as accuracy, precision and recall and F1-score to rate the prowess of the proposed model in taking part in the classifying process of e-waste components. From the analysis, the proposed DCNN model reached classification accuracy over 95.9% on all the categories suggesting the model can be used as a tool to sort e-waste automatically. The paper also considers the issues of implementing such technology in the current recycling systems and the overall advantages, which will be the increase of the sorting efficiency, decreased labor expenses, as well as elevated recovery rates of materials demanded in the market.