返回
An Improved Deep Learning Network for Spray Detection
DOI:10.1002/tee.70161.png)
摘要
En 中文
疏水性是表征复合绝缘子老化状态的一个重要指标。由于喷雾眩光干扰,常规计算机视觉方法在复合绝缘子现场疏水性检测中面临巨大挑战。为解决此挑战,本文提出了一种新型双流语义分割网络DelightNet,该网络分别从正常喷雾图像和喷雾图像的细节层进行学习,以准确检测和跟踪绝缘子上的液滴,并改进语义分割网络以减轻眩光干扰的影响。进行了多项实验来评估DelightNet的准确性,结果表明DelightNet的平均-score和平均mIoU高于多种经典深度学习网络:所提出的DelightNet实现了0.725的平均mIoU和0.809的平均-score,分别比当前最佳方法提高了2.4%和1.4%。(c) 2025 Institute of Electrical Engineers of Japan and Wiley Periodicals LLC.
Keyword:
composite insulators
semantic segmentation
low-frequency filtering
model parallelism
期刊
IF:
1.1
论文数:
251
被引数:
1.8K
机构
暂无机构信息
引用论文
A Smart Droplet Detection Approach With Vision Sensing Technique for Agricultural Aviation Application
IEEE SENSORS JOURNAL
IF4.5
Rethinking Semantic Segmentation from a Sequence-to-Sequence Perspective with Transformers从序列到序列的角度重新思考语义分割
Evaluation of Long-Term Boiling Water Immersion as Artificial Aging Test of Silicone Rubber Insulators长期沸水浸泡作为硅橡胶绝缘子人工老化试验的评估
Silicon
IF0
DropTrack-Automatic droplet tracking with YOLOv5 and DeepSORT for microfluidic applications
PHYSICS OF FLUIDS
IF4.3

