基于改进Yolov5s的输电线路防外力破坏行为检测识别

Detection and Identification of Transmission Line Damage Prevention Behavior Based on Improved Yolov5s

摘要:

电力系统的安全对于整个能源传输过程至关重要。针对输电线路下超大工程车辆和烟火为主要的外力破坏行为，对单阶段目标检测算法YOLOv5s进行改进，首先针对输电线路多雨雾烟尘等工作环境，引入限制对比度自适应直方图均衡算法CLAHE对图片进行去雾处理，提升图片对比度；针对检测目标距离较远的问题，在YOLOv5s网络的基础上添加CA注意力机制，提升了模型对目标的定位能力；将原网络中的最邻近差值采样方式替换为轻量级通用上采样算子CARAFE，更好地捕捉特征图的同时引入较小的参数量；最后在网络的特征融合层，使用具有通道混洗思想的GSConv卷积模块代替标准卷积模块，减小模型参数量，再利用slim_neck特征融合结构，强化目标关注度，达到减小模型参数量同时提升检测精度的效果。实验结果表明：改进后的YOLOv5s网络，mAP提升了4.4%，参数量减小了3.4%，权重模型内存减小了2.7%，证明了算法的有效性。

Abstract:

The safety of the power system is crucial for the entire energy transmission process. Aiming at the main external force destruction behavior of super large engineering vehicles and fireworks under the transmission line, the single-stage target detection algorithm YOOv5s is improved. First, aiming at the working environment of the transmission line with heavy rain, fog and dust, the restricted contrast adaptive histogram equalization equalization algorithm CLAHE is introduced to defog the image to improve the image contrast; In response to the problem of detecting targets with long distances, a CA attention mechanism was added to the YOLOv5s network to enhance the model's ability to locate targets; Replace the nearest neighbor difference sampling method in the original network with the lightweight universal upsampling operator CARAFE, which better captures feature maps while introducing smaller parameter quantities; Finally, in the feature fusion layer of the network, a GSConv convolution module with channel shuffling idea is used to replace the standard convolution module, reducing the number of model parameters, and then utilizing Slim_ Neck feature fusion structure enhances target attention, achieving the effect of reducing model parameters while improving detection accuracy. The experimental results show that the improved YOLOv5s network improves mAP by 4.4%, reduces parameter count by 3.4%, and reduces weight model memory by 2.7%, proving the effectiveness of the algorithm.