Abstract
Using UAV to accurately and quickly realize the three-dimensional modeling of power tower is an effective measure to promote the construction of smart grid. A tower target recognition algorithm based on yolov5 is presented to solve the problem of inaccurate tower 3D perception recognition by UAV. By improving the internal structure of the residual block to reduce the impact of steel intersection on the identification results and improve the accuracy of the algorithm, the exponential linear unit function is used as the activation function to accelerate the convergence speed of the network and improve the robustness of the algorithm, so as to realize the rapid and accurate identification of key points of power towers. By comparing the fast R-CNN and EfficientDet target detection methods through experiments, the improved algorithm has improved the recognition accuracy to a certain extent. The model maintains the lightweight characteristics of yolov5 and has a good prospect in application deployment.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Redmon J, Farhadi A (2017) YOLO9000: better, faster, stronger. In: IEEE conference on computer vision and pattern recognition. IEEE, Hawaii, USA, pp 6517−6525
Krizhevsky A, Sutskever I, Hinton GE (2017) ImageNet classification with deep convolutional neural networks. Commun ACM 60(6):84–90
Lin TY, Dollar P, Girshick R et al (2017) Feature pyramid networks for object detection. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 936–944
Clevert D-A, Unterthiner T, Hochreiter S (2016) Fast and accurate deep network learning by exponential linear units (ELUs). arXiv preprint arXiv:1511.07289v5
Redmon J, Farhadi A (2021) Yolov3: an incremental improvement [EB/OL]
Nordeng IE, Hasan A, Olsen D et al (2017) DEBC detection with deep learning. In: Scandinavian conference on image analysis. Springer, Troms, Norway, pp 248−259
Bochkovskiy A, Wang CY, Liao HYM (2021) Yolov4: optimal speed and accuracy of object detection [EB/OL]
Wei H, Zhang Y, Wang B et al (2020) X-linenet: detecting aircraft in remote sensing images by a pair of intersecting line segments. IEEE Trans Geosci Remote Sens
Paszke A, Gross S, Massa F et al (2019) PyTorch: an imperative style, high-performance deep learning library
Loshchilov I, Hutter F (2021) SGDR: stochastic gradient descent with warm restarts [EB/OL]
Liu Y, Ma Z, Liu X et al (2019) Privacy-preserving object detection for medical images with faster R-CNN. IEEE Trans Inf Forensic Secur (99):1
Jaeger PF, Kohl SAA, Bickelhaupt S et al (2020) Retina U-Net: embarrassingly simple exploitation of segmentation supervision for medical object detection. Mach Learn Health Workshop 171–183
Acknowledgements
This work was supported by the State Grid Xuzhou Power Supply Company, Xuzhou, Jiangsu, China (Research on UAV high-efficiency and fully autonomous power inspection technology based on real-time 3D spatial sensing and intelligent photography technology).
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.
About this paper
Cite this paper
Zhao, C., Cui, Y., Ding, Z., Cao, C. (2023). Key Point Detection of Power Tower Based on Improved Yolov5. In: Nayak, R.K., Pradhan, M.K., Mandal, A., Davim, J.P. (eds) Recent Advances in Materials and Manufacturing Technology. ICAMMT 2022. Lecture Notes in Mechanical Engineering. Springer, Singapore. https://doi.org/10.1007/978-981-99-2921-4_70
Download citation
DOI: https://doi.org/10.1007/978-981-99-2921-4_70
Published:
Publisher Name: Springer, Singapore
Print ISBN: 978-981-99-2920-7
Online ISBN: 978-981-99-2921-4
eBook Packages: EngineeringEngineering (R0)