Enhancing trash detection performance with YOLOv11n architecture using convolutional block attention module (CBAM)

The increasing volume of global waste demands more efficient management solu-Tions, particularly in waste classification and detection processes. In this context, intelligent waste management plays a crucial role in the development of smart cities, where technologies such as computer vision and artificial intelligence signifi-cantly enhance efficiency and support environmental sustainability. This study proposes the development of a computer vision-based waste detection model using YOLOv11n, modified by integrating the Convolutional Block Attention Module (CBAM) to strengthen feature extraction and improve the model's focus on important object regions. The dataset used is TrashNet, consisting of six waste categories: cardboard, glass, metal, paper, plastic, and trash, with data augmenta-Tion and oversampling techniques applied to address class imbalance. Experi-mental results show that the YOLOv11n + CBAM model achieved the best performance with Precision of 0.899, Recall of 0.916, mAP of 0.951, and mAP50-95 of 0.814, outperforming other YOLOv11n variants. The integration of the CBAM module effectively improves detection accuracy without significantly in-creasing computational complexity, making the model more efficient and accurate for supporting intelligent, computer vision-based waste management systems while contributing to the development of sustainable and environmentally friendly smart city infrastructure.