Waste classification is considered a major concern worldwide, as population growth directly contributes to the increase of waste generation. If not managed properly, the increase in waste generation can result in severe environmental impacts, including water pollution and health hazards Therefore, developing an automated solution for waste classification and detection has become significant. This study, Improves YOLOv8s model to address these challenges. The model was trained on Kaggle “(Garbage classification dataset. Accessed 2025.)” dataset that went through preprocessing steps such as auto-orientation and resizing techniques to enhance the model performance and data augmentation to increase the dataset diversity. This proposed model demonstrates strong performance, achieving an accuracy of 66.67%, MAP@50 of 92.19%, a precision of 96%, a recall of 97%, and a F1 score of 86%. Theses result highlights the model potentials for deployment in smart waste management systems to contribute to efficient, automated waste segregation and sustainable environment.