Scholarlink

<p>For the insulator discharge severity assessment at the line inspection site using edge-end computing equipment and UV cameras, this paper proposes an improved assessment algorithm based on the YOLOv8 algorithm. Firstly, LDConv is introduced to replace the convolution of the backbone network part of the network feature extraction, which effectively realizes the enhancement of the feature extraction ability of the algorithm in the case of model lightweighting; and then ACMix attention mechanism is introduced, which realizes better focusing of the model on the target with a very small performance loss; and finally, Shape-IoU is introduced to replace the loss function of the CIoU, which effectively improve the detection accuracy of the algorithm. The experimental results show that compared with the original YOLOv8, the RDIDSNet algorithm proposed in this paper achieves a detection speed of 61 Frames/s while realizing a detection accuracy of 78.1%, which can satisfy the demand for fast and accurate assessment of insulator discharge severity on edge devices.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Insulator discharge severity assessment algorithm based on RDIDSNet

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年4月8日

查看全文
  • Cheng Chi,
  • Li Keyu,
  • Yanhui Meng,
  • Yang Yang,
  • JiNing Zhao,
  • Shaotong Pei,
  • Haosen Sun

摘要

For the insulator discharge severity assessment at the line inspection site using edge-end computing equipment and UV cameras, this paper proposes an improved assessment algorithm based on the YOLOv8 algorithm. Firstly, LDConv is introduced to replace the convolution of the backbone network part of the network feature extraction, which effectively realizes the enhancement of the feature extraction ability of the algorithm in the case of model lightweighting; and then ACMix attention mechanism is introduced, which realizes better focusing of the model on the target with a very small performance loss; and finally, Shape-IoU is introduced to replace the loss function of the CIoU, which effectively improve the detection accuracy of the algorithm. The experimental results show that compared with the original YOLOv8, the RDIDSNet algorithm proposed in this paper achieves a detection speed of 61 Frames/s while realizing a detection accuracy of 78.1%, which can satisfy the demand for fast and accurate assessment of insulator discharge severity on edge devices.