改进YOLOv11模型在高压输电线路巡检中的应用

Improving the Application of YOLOv11 Model in High-Voltage Transmission Line Inspection

摘要: 针对高压输电线路巡检中传统人工方式效率低、漏检率高，以及现有目标检测算法对小目标缺陷检测精度不足、复杂背景抗干扰能力弱的问题，本文提出一种基于YOLOv11的改进巡检图像识别算法。设计自适应光照调整组合数据增强策略，解决巡检图像数据多样性不足、光照畸变影响检测的核心问题；在Neck层引入卷积块注意力模块与双向特征金字塔网络，强化缺陷区域特征提取与多尺度特征融合能力；优化检测头锚框尺寸与损失函数，采用CIoU损失替代原始损失函数以提升小目标定位精度。实验结果表明：改进算法的mAP达到96.5%，较原始YOLOv11提升4.3个百分点，FPS保持45 frame/s。

Abstract: To address the issues of low efficiency and high omission rates in traditional manual inspections of high-voltage transmission lines, as well as the insufficient detection accuracy for small target defects and weak anti-interference capability in complex backgrounds by existing object detection algorithms, an improved inspection image recognition algorithm based on YOLOv11 is proposed in this paper. An adaptive lighting adjustment combined with a data augmentation strategy is designed to resolve the core problems of insufficient data diversity in inspection images and the impact of lighting distortion on detection. A convolutional block attention module and bidirectional feature pyramid network are introduced in the Neck layer to enhance defect region feature extraction and multi-scale feature fusion capabilities. The detection head anchor box size and loss function are optimized, with the CIoU loss replacing the original loss function to improve small target localization accuracy. Experimental results show that the improved algorithm achieves an mAP of 96.5%, a 4.3 percentage point increase compared to the original YOLOv11, while maintaining an FPS of 45 frame/s.