WASTE CLASSIFICATION USING CONVOLUTIONAL NEURAL NETWORKS

Abstract

This study proposes a deep learning-based waste classification system using the MobileNetV2 architecture with transfer learning for efficient and accurate segregation. The model automatically classifies waste into three main categories—Recyclable, Organic, and Non-Recyclable—achieving a validation accuracy of up to 90%. Techniques such as data augmentation, fine-tuning, and early stopping were applied to enhance generalization and reduce overfitting. An interactive Gradio interface enables real-time image uploads, predictions, and recycling suggestions, making the system practical for community and municipal waste management. Designed to operate efficiently in real-time and resource-limited environments, the proposed approach supports automation in waste sorting and promotes environmental sustainability. The system contributes to smart city initiatives and aligns with the United Nations Sustainable Development Goals (SDG 12: Responsible Consumption and Production, and SDG 13: Climate Action).