Scalable Deep Learning Techniques for Automated Waste Segregation in Smart City Environments

Effective waste management is an important aspect of realising the United Nations' Sustainable Development Goals, especially SDG 11 (Sustainable Cities and Communities) and SDG 12 (Responsible Consumption and Production). In this research, an automated waste classification using deep learning techniques is developed with a focus on further development of the segregation and management processes. Here, two models were implemented: CNN and VGG16 with data augmentation to enhance classification ability. The CNN model outperformed the other model in the test with an accuracy of 95.15% and a loss of 0.1336 on the validation set, while the VGG16 model returned an accuracy of 94.43% and a loss of 0.309. Comparing this in light of its simpler structure with more effective feature extraction and generalization, the model seems more suited for the waste classification task. It segregates waste into categories such as recyclables and organics, aligning it with SDG targets through the promotion of sustainable waste management and environmental impact reduction. Future research will be designed toward real-time implementation and an expansion of the dataset to further drive the effectiveness of the system in support of sustainable urban development.