SMW-YOLO: Advancing Small Object Detection in Remote Sensing based on YOLOv11

This paper addresses the challenges of small target detection in remote sensing images, such as few target pixels, limited features, and complex backgrounds, by proposing an improved YOLOv11 model. The main improvements include: (1) designing the C3K2_StarsBlock module, which enhances the feature extraction of small targets through a star-shaped structure and feature aggregation; (2) proposing the MSAA module that integrates multi-scale features and attention mechanisms, combined with residual connections to improve model robustness; (3) using decoupled detection heads and the WIoU loss function to optimize classification and localization tasks. Experiments on the NWPU VHR-10 and DOTA datasets show that the proposed method improves mAP by 2.7 % and 2.1 % respectively, verifying its effectiveness.