This paper addresses the challenges of small object detection in UAV aerial imagery by proposing an improved YOLOv9 object detection model. The core innovations are twofold: first, the introduction of a Global Attention Mechanism (GAM), which enhances the model's perception of small object features through channel and spatial dual-path attention processing; second, the adoption of a Bidirectional Feature Pyramid Network (BiFPN), which implements bidirectional feature fusion from top-down and bottom-up perspectives, effectively improving the interaction efficiency of features at different scales. Experiments on the VisDrone2019 dataset demonstrate that, compared to the baseline YOLOv9, the proposed model improves the mAP@0.5 metric by 1.6%, while reducing the parameter count by 2.8×10^6. Visual comparisons show that the model exhibits superior detection capabilities in complex environments such as aerial multi-scale small objects, dense crowds, and night scenes, effectively addressing the problems of missed detections and false detections of small objects.