Automated Household E-Waste Recognition and Classification Using Faster R-CNN and Modified ResNet50

Abstract

Electronic waste (E-waste) has emerged as a major environmental challenge due to the rapid growth in electronic device consumption and the improper disposal of end-of-life equipment. Traditional E-waste collection and segregation processes are often labor-intensive, inefficient, and expose workers to hazardous materials. To address these challenges, this study proposes a vision-enabled E-waste management system that integrates Faster Region-Based Convolutional Neural Network (Faster R-CNN) and a Modified ResNet50 model for automated detection and classification of household electronic waste. The proposed framework is designed to support waste collection operations by identifying E-waste items through a vehicle-mounted camera system. Image preprocessing techniques, including resizing, normalization, and data augmentation, were employed to improve model robustness and classification performance. Faster R-CNN was utilized for object detection, while the Modified ResNet50 model was developed using transfer learning for accurate classification of multiple E-waste categories. The performance of the proposed model was evaluated and compared with AlexNet and VGG16 architectures using accuracy, precision, recall, and F1-score metrics. Experimental results demonstrated that the proposed Modified ResNet50 model achieved a classification accuracy of 97%, outperforming the comparative models. The proposed system provides an efficient and scalable solution for automated E-waste identification, reducing manual effort and supporting sustainable recycling practices in smart-city waste management applications.