An improved insulator self-explosion detection method based on group-level pruning for the YOLOv7-tiny algorithm

被引:1
作者
You, Xilai [1 ]
Ma, Jianqiao [1 ]
Yang, Guangze [2 ]
机构
[1] Lanzhou Jiaotong Univ, Sch Automat & Elect Engn, Dept Power Engn, Lanzhou 730070, Peoples R China
[2] State Grid Shandong Elect Power Co, Taian Power Supply Co, Tai An 271000, Shandong, Peoples R China
关键词
Insulators; YOLOv7; Model pruning; Object detection; Attention mechanism;
D O I
10.1007/s11554-024-01571-0
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
With the construction of intelligent grids, unmanned aerial vehicle have been widely employed to inspect transmission lines. The inspection process generates a large amount of data, which requires a lightweight model to reduce computational overhead. Here, we propose an improved model based on YOLOv7-tiny with group-level pruning to reduce the model size, which achieves a balance between detection accuracy and speed. Firstly, we replace the activation function with the Funnel activation function to optimize the activation domain dynamically. Second, we introduce a lightweight DFC attention mechanism to enhance the ability of backbone to extract long-range features. Finally, we use adaptively spatial feature fusion network to reduce semantic degradation during feature fusion. We group the parameters according to their dependencies and use a consistent sparse approach to obtain parameter importance. The redundant parameter groups were pruned to achieve model light-weighting. Experimental results show that the improved model achieves 95.6% detection accuracy after pruning. Compared with YOLOv7-tiny, the computational complexity is reduced by 53% and the processing speed is increased by 48.1% to 73 frames per second.
引用
收藏
页数:13
相关论文
共 50 条
[41]   Mask wearing detection algorithm based on improved YOLOv7 [J].
Luo, Fang ;
Zhang, Yin ;
Xu, Lunhui ;
Zhang, Zhiliang ;
Li, Ming ;
Zhang, Weixiong .
MEASUREMENT & CONTROL, 2024, 57 (06) :751-762
[42]   Pedestrian Fall Detection Algorithm Based on Improved YOLOv7 [J].
Wang, Fei ;
Zhang, Yunchu ;
Zhang, Xinyi ;
Liu, Yiming .
NEURAL COMPUTING FOR ADVANCED APPLICATIONS, NCAA 2024, PT I, 2025, 2181 :437-448
[43]   Multi-Objects Recognition and Self-Explosion Defect Detection Method for Insulators Based on Lightweight GhostNet-YOLOV4 Model Deployed Onboard UAV [J].
Zhang, Shihai ;
Qu, Chongnian ;
Ru, Chengyin ;
Wang, Xingcai ;
Li, Zhenwei .
IEEE ACCESS, 2023, 11 :39713-39725
[44]   Fully Automatic and Precisely Woven Fabric Defect Detection Using Improved YOLOv7-Tiny Model Utilizing Enhanced Residual Convolutional Network [J].
Barman, Jagadish ;
Kuo, Chung-Feng Jeffrey .
FIBERS AND POLYMERS, 2025, 26 (01) :353-368
[45]   Improved Cherry Detection Method at Night Based on YOLOv7: YOLOv7-Cherry [J].
Gai, Rongli ;
Kong, Xiangzhou ;
Qin, Shan ;
Wei, Kai .
Computer Engineering and Applications, 2024, 60 (21) :315-323
[46]   YOLOv7-GCM: a detection algorithm for creek waste based on improved YOLOv7 model [J].
Qin, Jianhua ;
Zhou, Honglan ;
Yi, Huaian ;
Ma, Luyao ;
Nie, Jianhan ;
Huang, Tingting .
PATTERN ANALYSIS AND APPLICATIONS, 2024, 27 (04)
[47]   A Ginkgo Detection Algorithm in Complex Environments Based on Improved YOLOv7 [J].
Ma, Chi ;
Guo, Qiang .
2024 9TH INTERNATIONAL CONFERENCE ON INTELLIGENT COMPUTING AND SIGNAL PROCESSING, ICSP, 2024, :155-158
[48]   A Steel Surface Defect Detection Algorithm Based on Improved YOLOv7 [J].
Mao, Yihai ;
Zhang, Hongyi ;
Gao, Xingen ;
Luan, Shen ;
Lin, Yuxing ;
Qi, Xuanhao .
PROCEEDINGS OF 2023 7TH INTERNATIONAL CONFERENCE ON ELECTRONIC INFORMATION TECHNOLOGY AND COMPUTER ENGINEERING, EITCE 2023, 2023, :1096-1101
[49]   Improved YOLOv7-based steel surface defect detection algorithm [J].
Xie, Yinghong ;
Yin, Biao ;
Han, Xiaowei ;
Hao, Yan .
MATHEMATICAL BIOSCIENCES AND ENGINEERING, 2024, 21 (01) :346-368
[50]   A insulator defect detection network based on improved YOLOv7 for UAV aerial images [J].
You, Xingyi ;
Zhao, Xiaohu .
MEASUREMENT, 2025, 253