Maintaining waste management is one of the important steps to achieve environmental sustainability. Often, traditional techniques of sorting and disposal are inadequate, resulting in increased landfill use, greenhouse gas emissions, and overconsumption of natural resources. The key idea of the litter sorting process is to utilize artificial intelligence power to automate the litter sorting process while assessing the environmental impact. In the proposed research work, waste has been classified into different categories through a deep learning model, ResNet50, with a validation accuracy of 96.78%. CO2 emission data show that plastic waste has been responsible for the highest emissions of 798 kg CO2e among all the waste types. Scenarios for reduction predicted that a 50% reduction in plastic waste could significantly reduce emissions in a system of this scale and effect. Overall, this study shows the applicability of AI technologies for improving waste management, cutting emissions, and helping to drive policy toward a more sustainable future. The findings provide actionable inputs to the government, enterprises, and environmental stakeholders for the development of innovative waste management strategies.