In recent years, sustainable waste management systems in urban contexts depend heavily on the effective sorting of garbage inside smart dustbins. This research presents a unique framework using sophisticated deep learning methodologies and image processing techniques for garbage identification and segmentation within smart dustbins. In order to identify various waste objects within the bin compartments, the suggested technique entails preprocessing steps for picture enhancement followed by object recognition using convolutional neural networks (CNNs). A segmentation network is then applied to fine-tune the identified areas and accurately define the boundaries of every trash item. An extensive dataset of garbage photos taken in a range of lighting and waste composition scenarios is used to assess the robustness and efficacy of the suggested framework. High recall rates and precision in correctly identifying and classifying waste objects are demonstrated by the experimental findings, which improve the effectiveness of waste sorting procedures in smart dustbins.