Improved YOLOv5 foreign object detection for transmission lines

Abstract

The traditional transmission line detection has the problems of low efficiency. To improve the performance, this paper proposes an improved you only look once version 5 (YOLOv5) transmission line foreign object detection algorithm. First, efficient channel attention (ECA) module is introduced in the backbone network for focusing the target features and improving the feature extraction capability of the network. Secondly, bilinear interpolation upsampling is introduced in the neck network to improve the model detection accuracy. Finally, by integrating the efficient intersection over union (EIoU) loss function and Soft non-maximum suppression (Soft NMS) algorithm, the convergence speed of the model is accelerated while the detection effect of the model is enhanced. Relative to the original algorithm, the improved algorithm reduces the number of parameters by 16.4%, increases the mean average precision (mAP)@0.5 by 3.9%, mAP@0.5: 0.95 by 6.3%, and increases the detection speed to 55.3 frames per second (FPS). The improved algorithm is able to improve the performance of the foreign object detection in transmission lines effectively.

References

[[1]] Zhang Q Y, Zhu T, Xiao S Z, et al.. Foreign object detection of high voltage transmission line based on improved YOLOv4 algorithm[J]. Applied science and technology, 2023, 50(4): 59-65 (in Chinese)

[[2]] Zhou H. . Research on detection methods of foreign objects in power transmission lines[D], 2021 Chongqing Chongqing University of Technology (in Chinese)

[[3]] Tang X X, Shen W, Zhu M, et al.. The foreign object detecting algorithm for transmission lines based on the improved YOLOv4[J]. Journal of Anhui University (Natural Science Edition), 2021, 45(5): 58-66 (in Chinese)

[[4]] Qi Y C, Wu X L, Zhao Z B, et al.. Bolt defect detection for aerial transmission lines using Faster R-CNN with an embedded dual attention mechanism[J]. Journal of image and graphics, 2021, 11: 2594-2604 (in Chinese)

[[5]] Yuan J, Zheng X, Peng L, et al.. Identification method of typical defects in transmission lines based on YOLOv5 object detection algorithm[J]. Energy reports, 2023, 9: 323-332,

[[6]] YU C, LIU Y, ZHANG W, et al. Foreign objects identification of transmission line based on improved YOLOv7[J]. IEEE access, 2023.

[[7]] LIU M, LI Z, LI Y C, et al. A method for transmission line defect edge intelligent inspection based on re-parameterized YOLOv5[J/OL]. High voltage engineering, [2023-07-28]. https://doi.org/10.13336/j.1003-6520.hve.202-20861. (in Chinese)

[[8]] Xu Z J, Zhang Q, Xu L. A traffic sign recognition method based on improved YOLOv5s-ghost network[J]. Journal of optoelectronics-laser, 2023, 34(01): 52-61 (in Chinese)

[[9]] Wang Q, Wu B, Zhu P, et al.. ECA-Net: efficient channel attention for deep convolutional neural networks[C]. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, June 13–19, 2020, Seattle, WA, USA, 2020 New York IEEE 11534-11542

[[10]] Srinivas A, Lin T Y, Parmar N, et al.. Bottleneck transformers for visual recognition[C]. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, June 20–25, 2021, Virtual, 2021 New York IEEE 16519-16529

[[11]] LIU Y, SHAO Z, HOFFMANN N. Global attention mechanism: retain information to enhance channel-spatial interactions[EB/OL]. (2021-12-10) [2023-07-28]. https://arxiv.org/abs/2112.05561.

[[12]] Zhang R, Zhang Y R. Improved location fingerprint position algorithm based on the combination of bilinear interpolation and K-means clustering algorithm[J]. Optical communication technology, 2022, 46(05): 45-49 (in Chinese)

[[13]] Wang J, Chen K, Xu R, et al.. CARAFE: content-aware reassembly of features[C]. Proceedings of the IEEE/CVF International Conference on Computer Vision, October 27–November 2, 2019, Seoul, Korea (South), 2019 New York IEEE 3007-3016

[[14]] Ma Q W, Liu G H. Surface defect detection method of strip steel based on lightweight YOLOv3[J]. Journal of optoelectronics-laser, 2023, 34(7): 743-751 (in Chinese)

[[15]] ZHANG Y F, REN W, ZHANG Z, et al. Focal and efficient IOU loss for accurate bounding box regression[EB/OL]. (2021-01-20) [2023-07-28]. https://arxiv.org/abs/2101.08158.

[[16]] GEVORGYAN Z. SIoU loss: more powerful learning for bounding box regression[EB/OL]. (2022-05-25) [2023-07-28]. https://arxiv.org/abs/2205.12740.

[[17]] Rezatofighi H, Tsoi N, Gwak J Y, et al.. Generalized intersection over union: a metric and a loss for bounding box regression[C]. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, June 15–20, 2019, Long Beach, USA, 2019 New York IEEE 658-666

[[18]] Wang C Y, Bochkovskiy A, Liao H Y M. YOLOv7: trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, July 18–22, 2023, Vancouver, Canada, 2023 New York IEEE 7464-7475

[[19]] Zhao L, Wang S Z, Shao M W, et al.. Improved YOLOv5-based bird’s nest defect detection method for transmission lines[J]. Electronic measurement technology, 2023, 3: 157-165 (in Chinese)