In this paper, an EF-YOLOv8 model based on YOLOv8 is proposed to solve the problem of detecting small target defects in power equipment. This model significantly improves the ability to identify subtle defects through innovative architectural adjustments. The core contribution lies in the design of the C2f-EF module, which combines multi-scale features with an enhanced attention mechanism to specifically improve the model's perception to detect small targets. The model also integrates the improved spatial pyramid pooling structure SPPCSPC-G and weighted bidirectional feature pyramid network (BiFPN), which further promotes accurate integration and extraction of multi-scale features and improves small target detection performance. Experiments show that EF-YOLOv8 has achieved remarkable results in small target defect detection. The mAP index has increased by 2.7% and 4.7%. At the same time, the model efficiency has been greatly improved. Parameters and computation were reduced by 12.5% and 10.9%, respectively. The detection speed is increased to 62 frames/s. This study provides a new efficient and accurate solution for defect detection in the field of power operation inspection, which is of great significance for improving the level of safety operation and maintenance of power grids.