The safe operation of power equipment is of paramount importance. However, power equipment may suffer various types of damage during long-term operation due to a variety of factors, leading to equipment failure or even serious power accidents. To address this issue, this paper proposes a transmission line defect detection model based on improved YOLOv8n. First, the feature extraction network of the algorithm is improved by using the MAB attention mechanism, which enhances the ability to extract relevant features. Then, Pconv is used to replace the first layer of Conv to enhance the underlying feature extraction and expand the receptive field. Subsequently, the connection between the trunk and neck is improved by borrowing the cross-layer connection from BiFPN, so that the model can fuse more positional and detailed information, and make the network better focus on the area where the small target is located. Finally, the SDB loss function is used instead of CIoU Loss to improve the model’s detection performance for small targets. Experimental results show that our proposed MP-YOLO model improves precision by 0.6%, recall by 4.2%, and mAP₅₀ by 2% compared to the standard YOLOv8n.