Smart Waste Management with Deep Learning based Classification

Accurate trash classification is essential for both environmental sustainability and efficient waste management. Since improper garbage disposal adds to pollution, effective classification methods are crucial for better recycling and waste management. In order to improve classification accuracy, this study introduces a deep learning-based waste classification system that combines MobileNetV2 and ResNet50V2. Using the TrashNet dataset, the system attains a validation accuracy of 94.44% by utilizing sophisticated picture augmentation techniques and fine-tuning the model. The approach helps automate recycling processes by efficiently classifying six different waste types: cardboard, glass, metal, paper, plastic, and miscellaneous rubbish. Real-time waste classification and smooth integration into automated waste management systems are made possible by the system's implementation as a web application using Streamlit to assure practical usage.