DL-Clean: CNN-Based Waste Classification for Smart City Infrastructure

Automated waste sorting is a tedious and, more importantly, an error-prone task. This study attempts to channel the potential of CNNs in fighting pollution to the automation of waste sorting. We trained a model that accepted images of waste and classified them into different types using CNN image recognition. We created a system and evaluated it using a rich dataset. The system achieved 90.42% accuracy and 89.16% validation accuracy during training. The model also achieved high performance in classifying waste with 92.91% precision, 92.19% recall, and 92.20% F1-score. The combination of CNNs and IoT waste management systems achieves greater efficiency in real time than other existing models. This enhances the resource management productivity and efficiency of automated sorting in smart cities.