Images obtained from drones offer more reliable and comprehensive information. However, equipment stability, perspective, and scale variations contribute to challenges in drone aerial images, which often exhibit characteristics like low resolution for small targets, complex lighting conditions, and multi-scale features. This study presents an enhanced model, WW-YOLO, derived from YOLOv8n, designed to improve detection efficiency and accuracy in resource-limited environments. First, the paper proposes an innovative feature-focus diffusion network at the neck of the network architecture. This novel approach combines high- and low-dimensional features by utilizing the Multi-dimensional Selective Integration Module. Secondly, to further strengthen the learning capacity for small targets and global features, a multi-branch feature extraction module, G2LM, is proposed. Finally, to further streamline the model, the Detail-Enhanced Shared Light-weight Detection Head is proposed, and wise-iou is employed as the loss function, significantly reducing computational resource consumption and enhancing detection efficiency. The experiment's findings indicate that in the public dataset VisDrone2019, WW-YOLO attained a 5.1% enhancement in mean average precision relative to the baseline model, concurrently decreasing the parameter count by 74%. WW-YOLO demonstrates superior performance in object detection from drone aerial imagery compared to other related methods.