Waste classification is a critical component of solid waste management, aimed at improving recycling efficiency and environmental sustainability. This paper presents a Convolutional Neural Network (CNN)-based approach for classifying waste into biodegradable and non-biodegradable categories. The empirical study explores the performance of two CNN architectures namely ResNet50 and VGG16. While ResNet50 was tested, its performance was suboptimal, leading to the selection of VGG16, which achieved superior accuracy in classifying waste. The system has been tailored for fast-growing cities like Bengaluru, India, which generates approximately 4500 metric tons of waste daily. The VGG16 model’s higher accuracy demonstrates its potential for automating waste segregation and supporting sustainable waste management practices. The research highlights the growing need for advanced technological solutions to manage urban waste efficiently and suggests that further optimization of deep learning models could improve accuracy and scalability. Additionally, the research provides insights into the challenges and opportunities in deploying CNN-based models in real-world waste management systems.