Ensuring the reliability and safety of electrical power systems requires the efficient detection of defects in high-voltage transmission line insulators, which play a critical role in electrical isolation and mechanical support. Environmental factors often lead to insulator defects, highlighting the need for accurate detection methods. This paper proposes an enhanced defect detection approach based on a lightweight neural network derived from the YOLOv11n architecture. Key innovations include a redesigned C3k2 module that incorporates multidimensional dynamic convolutions (ODConv) for improved feature extraction, the introduction of Slimneck to reduce model complexity and computational cost, and the application of the WIoU loss function to optimize anchor box handling and to accelerate convergence. Experimental results demonstrate that the proposed method outperforms existing models like YOLOv8 and YOLOv10 in precision, recall, and mean average precision (mAP), while maintaining low computational complexity. This approach provides a promising solution for real-time, high-accuracy insulator defect detection, enhancing the safety and reliability of power transmission systems.

UAV-based insulator defect detection faces persistent challenges of multi-scale defect variation, background clutter, small-target missed detections and redundant detection-head computation. This paper proposes BEAD-Net, a real-time insulator defect detection network built upon YOLOv11n with four targeted improvements. First, an Hourglass Symmetric Residual Attention (HSRA) module replaces the standard bottleneck components within C3k2, expanding the multi-scale receptive field and suppressing background interference via a symmetric hourglass dilation schedule and channel attention recalibration. Second, a Bidirectional Diffusion Feature Pyramid Network (BDFPN) built upon the Group-wise Selective Feature Integrator (GSFI) employs group-wise adaptive gating and two-level bidirectional propagation to mitigate semantic dilution during cross-scale fusion. Third, a Task-Decoupled Asymmetric Detection Head (TDAH) concentrates spatial modeling in the regression branch while simplifying classification to lightweight $1\times 1$ operations, reducing parameter redundancy and alleviating inter-task gradient conflict. Finally, Focaler-PIoU2 integrates linear interval mapping with a normalized corner distance penalty to improve boundary regression for slender insulator structures. On the IDID dataset, BEAD-Net achieves mAP@0.5 of 83.98% and mAP@0.5:0.95 of 63.82% at 275.04 FPS with 2.49 M parameters, outperforming the baseline YOLOv11n by 2.83 and 1.55 percentage points and surpassing all compared state-of-the-art methods. Additional validation on the CPLID benchmark shows that BEAD-Net remains the top performer, confirming that the proposed improvements are not specific to a single dataset.

In the area of transmission line inspection, one of the prominent areas of research has been to unite Unmanned Aerial Vehicles (UAVs) with neural network object detection algorithms. This area of research is challenging because of high computational resource consumption and poor infrared detection capabilities. In this study we propose an infrared image detection algorithm, named IISD-YOLO, using a modified version of the YOLOv11 network, to detect infrared transmission line insulator strings. Firstly, the original object detection layer was removed and replaced with the ShuffleNetv2 network to achieve the goal of a lightweight model; subsequently, based on the original feature extraction module C3k2, the Manhattan Self-Attention (MaSA) mechanism was introduced to design a new feature extraction module, C3k2-MaSA, which enhances the feature extraction capability for infrared objects; finally, the bidirectional feature pyramid network (Bi-FPN) is used to replace the original feature fusion module, enhancing the network’s ability to process and fuse information at different scales. The comparative experiments show that compared with the mainstream YOLO models, IISD-YOLO has improved by 4.5, 6.1, and 4.8 percentage points respectively on mAP@50 over YOLOv5, YOLOv8, and YOLOv10; furthermore, this model outperforms advanced models including YOLO-CIR, FA-YOLO, YOFIR, and RT-DETR, with improvements of 2.9, 9.1, 5.0, and 1.1 percentage points respectively on mAP@50. The ablation study shows that each improvement effectively enhances the overall performance. Compared with the original YOLOv11, the IISD-YOLO has increased its mAP@50 by 3.5 percentage points, while reducing the number of Params by 1.1 million and the computational GFLOPs by 2 G. These results confirm the superior performance of IISD-YOLO in infrared insulator string detection.