This paper proposes a small target detection algorithm for UAV images based on an improved YOLOv8 model, termed YOLOv8-MPEB. The algorithm addresses the challenges of detecting small targets in UAV aerial images, including large-scale changes, small target sizes, complex scenes, and variable external factors. The backbone network is replaced with MobileNetV3 to reduce model parameters and computational complexity while enhancing inference speed. A dedicated small target detection layer is designed to optimize feature extraction for multi-scale targets. The Efficient Multi-Scale Attention (EMA) mechanism is integrated into the Convolution to Feature (C2f) module to enhance feature extraction and suppress redundant features. A bidirectional feature pyramid network (BiFPN) is used in the Neck segment to improve detection accuracy and generalization. The algorithm achieves a mean Average Precision (mAP) of 91.9% on a custom-made helmet and reflective clothing dataset, with a parameter count of 7.39 M and a model size of 14.5 MB. Compared to standard YOLOv8 models, the algorithm improves average accuracy by 2.2 percentage points, reduces model parameters by 34%, and decreases model size by 32%. It outperforms other prevalent detection algorithms in terms of accuracy and speed. The algorithm is evaluated through ablation experiments and comparative experiments, demonstrating its effectiveness in detecting small targets in UAV images. The improved YOLOv8 model shows superior performance in multi-scale small-target detection and generalization ability for UAV images compared to YOLOv8s. The algorithm effectively reduces leakage and false detection in UAV images, although challenges remain in detecting tiny, aggregated, and similar targets. Future work will focus on optimizing the multiscale feature pyramid strategy and localization loss function to improve algorithm accuracy and model performance in scenarios with small target aggregations.This paper proposes a small target detection algorithm for UAV images based on an improved YOLOv8 model, termed YOLOv8-MPEB. The algorithm addresses the challenges of detecting small targets in UAV aerial images, including large-scale changes, small target sizes, complex scenes, and variable external factors. The backbone network is replaced with MobileNetV3 to reduce model parameters and computational complexity while enhancing inference speed. A dedicated small target detection layer is designed to optimize feature extraction for multi-scale targets. The Efficient Multi-Scale Attention (EMA) mechanism is integrated into the Convolution to Feature (C2f) module to enhance feature extraction and suppress redundant features. A bidirectional feature pyramid network (BiFPN) is used in the Neck segment to improve detection accuracy and generalization. The algorithm achieves a mean Average Precision (mAP) of 91.9% on a custom-made helmet and reflective clothing dataset, with a parameter count of 7.39 M and a model size of 14.5 MB. Compared to standard YOLOv8 models, the algorithm improves average accuracy by 2.2 percentage points, reduces model parameters by 34%, and decreases model size by 32%. It outperforms other prevalent detection algorithms in terms of accuracy and speed. The algorithm is evaluated through ablation experiments and comparative experiments, demonstrating its effectiveness in detecting small targets in UAV images. The improved YOLOv8 model shows superior performance in multi-scale small-target detection and generalization ability for UAV images compared to YOLOv8s. The algorithm effectively reduces leakage and false detection in UAV images, although challenges remain in detecting tiny, aggregated, and similar targets. Future work will focus on optimizing the multiscale feature pyramid strategy and localization loss function to improve algorithm accuracy and model performance in scenarios with small target aggregations.