SmartSeg AI: A CNN-Based Intelligent Waste Classification Framework with Knowledge-Based Waste Decision Engine

Abstract

Improper waste segregation is one of the major environmental challenges due to the increase in the generation of municipal solid waste and limitation of manual sorting methods. SmartSeg AI presents a smart waste sorting system framework that involves a convolutional neural network with a Knowledge-Based Waste Decision Engine (KB-WDE) for effective waste segregation. This work proposes a system that classifies waste images into 12 classes and maps the predicted object class to its waste category, recyclability status, recommended disposal destination, and disposal guidance. To improve the real-time usability of the classification result, the KB-WDE converts the predicted object class to its waste category, recyclability, recommended disposal bin, and disposal guidance. This model was trained and evaluated using a waste image dataset containing 15,515 labelled images across all 12 waste classes. Our framework is suitable for future integration into automated smart waste segregation systems, as it provides actionable waste management recommendations. The system also allows for easy integration with IoT-enabled smart bins and robotic waste sorting platforms.