Published in Scientific Papers. Series E. Land Reclamation, Earth Observation & Surveying, Environmental Engineering, Vol. XIII
Written by Dan Constantin PUCHIANU
The integration of convolutional neural networks and modern web applications can significantly improve the efficiency of recycling processes. Accurate, rapid identification and separation of waste of various types reduces contamination by marking a process essential to the efficiency of the recycling industry. In this study, a modern approach for classifying recyclable waste using deep-learning techniques based on convolutional neural networks and integrated into a web application developed using ReactJS is presented. Leveraging the features of advanced deep-learning models and modern web interfaces, the present study aims to make a substantial contribution to the field of efficient waste management and environmental protection. Neural network architectures, trained and evaluated on a carefully annotated dataset, demonstrated very good accuracy values outperforming classical state-of-the-art models. Integrating these models with modern web technologies built a web application with an intuitive user interface for real-time classification of waste types, providing immediate feedback. In the same framework, implementations with web technologies also provide educational resources regarding recycling practices and the impact of waste on the environment. The impact on the environment is considerable because the development of such established technologies can reduce the amount of waste managed improperly, improving the recycling rate. Future research can explore optimizing the models and techniques presented in these studies, expanding the dataset, and developing the application to support good sustainability practices.
[Read full article] [Citation]
