AI-Driven Waste Classification Model for Recycling Applications using YOLOv11

In developing countries waste management is a big issue, particularly in urban areas where rapid population growth and industrialization generate considerable amounts of waste. While Tons of waste is generated on daily basis, only a fraction of it gets recycled. A major factor here elements mixture. Nation wide Missons are running to collect wet and dry waste separately, where wet goes for composting and dry comes for recycling. This dry waste is segregated by human which is very expensive considering time, efforts and health issues and is prone to errors. It is said that where humans can’t reach machines can. In this field also, research have proved that machine can work effectively in waste management with minimal cost and accuracy so this study will used latest Artificial Intelligence (AI) model known as YOLOv11 i.e. You Only Look Once Version 11 for object detection and classification. Dataset used here is named as "Garbage Classification" available on Kaggle. Dataset contains approximately 20,000 images, which were standardized to 640 by 640 pixels and labels were generated manually. Model was trained on ~16000 images of 10 categories which were battery, biological, cardboard, cloth, glass, metal, paper, plastic, shoes and thermocol. Over 50 Epochs, Mean Average Precision reached to ~0.99 and accuracy of model to predicted items over 10 categories was above 85%. With further training this system not only reduces human effort but also enhances the efficiency and scalability for industrial waste management processes.