政大機構典藏-National Chengchi University Institutional Repository(NCCUR):Item 140.119/136960
English  |  正體中文  |  简体中文  |  Post-Print筆數 : 27 |  全文筆數/總筆數 : 113451/144438 (79%)
造訪人次 : 51340390      線上人數 : 846
RC Version 6.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
搜尋範圍 查詢小技巧：
  • 您可在西文檢索詞彙前後加上"雙引號"，以獲取較精準的檢索結果
  • 若欲以作者姓名搜尋，建議至進階搜尋限定作者欄位，可獲得較完整資料
    •  進階搜尋
    政大機構典藏 > 資訊學院 > 資訊科學系 > 學位論文 >  Item 140.119/136960


    請使用永久網址來引用或連結此文件: https://nccur.lib.nccu.edu.tw/handle/140.119/136960


    題名: 基於深度學習框架之電塔與絕緣子檢測
    Power Tower and Insulator Detection Based on Deep Learning Framework
    作者: 陳冠增
    Chen, Kuan-Tseng
    貢獻者: 廖文宏
    Liao, Wen-Hung
    陳冠增
    Chen, Kuan-Tseng
    關鍵詞: 電腦視覺
    物件偵測
    深度學習
    Computer vision
    Object detection
    Deep learning
    日期: 2021
    上傳時間: 2021-09-02 16:53:35 (UTC+8)
    摘要: 深度學習技術改變了電腦視覺領域的理論與實作架構，近年來已經被廣泛地應用，本論文主要的目標是將這些新興的技術道入電塔與絕緣子的自動偵測任務，除了蒐集更多視角的電塔資料，也針對不同的網路架構、骨幹與優化方式，進行全面式的分析比較，提升對電塔與絕緣子的辨識準確率，以期運用於無人機的安全檢測。
    本論文共實驗CascadeRCNN、FasterRCNN、Retinanet、Fcos、DetectoRS、YOLO V3與YOLO V4七種架構，搭配Resnet50、Resnet101、Regnet、Resnest、Resnext與Res2net六個骨架模型，測試七個優化方法，分別為Libra-RCNN、Soft-nms、Weight Standardization、OHEM、DCNv2、Multi-scale training與Multi-scale testing，以尋找最佳組合。
    在電塔的檢測中，本論文發現最佳組合為CascadeRCNN架構，搭配Resnext、Weight Standardization與Multi-scale training，取得AP = 91.06%、AR = 93.72%以及F1 Score = 92.37%。而絕緣子的最佳組合為DetectoRS架構，搭配Regnet、Weight Standardization與Multi-scale training，取得AP = 95.66%、AR = 97.92%以及F1 Score = 96.78%，最終的準確率與現有方法相比，都獲得相當程度的提升。
    Deep learning technology has brought forth a paradigm shift in the field of computer vision in both theoretical and practical aspects. It has been widely incorporated in many applications recently. The main objective of this research is to introduce these emerging technologies into the automatic detection of power towers and insulators. Toward this goal, we investigate the combination of different network architectures, backbones and optimization methods to compare and analyze their performance in the identification of power towers and insulators, with an expectation to apply them to the safety inspection using UAVs.
    In this thesis, we tested seven architectures, including CascadeRCNN, FasterRCNN, Retinanet, Fcos, DetectoRS, YOLO V3 and YOLO V4, with six backbone models, i.e., Resnet50, Resnet101, Regnet, Resnest, Resnext and Res2net, and seven optimization methods, namely Libra- RCNN, Soft-nms, Weight Standardization, OHEM, DCNv2, Multi-scale training and Multi-scale testing to search for the best combination.
    In the detection of power towers, we found that the best combination is CascadeRCNN with Resnext, Weight Standardization and Multi-scale training, achieving results of AP = 91.06%, AR = 93.72% and F1 Score = 92.37%. The best combination for insulator detection is DetectoRS, combined with Regnet, Weight Standardization and Multi-scale training, obtaining results of AP = 95.66%, AR = 97.92% and F1 Score = 96.78%. The final accuracy shows notable improvements over existing methods.
    參考文獻: 1.ImageNet Large Scale Visual Recognition Competition (ILSVRC). http://www.image-net.org/challenges/LSVRC/, last visited on Dec 2018.
    2.ImageNet. http://www.image-net.org/, last visited on Dec 2018.
    3.C. Szegedy, W. Zaremba, I. Sutskever, J. Bruna, D. Erhan, I. Goodfellow, R. Fergus, Intriguing properties of neural networks, arXiv preprint arXiv:1312.6199, 2014.
    4.Warren S. McCulloch, Walter H. Pitts. A logical calculus of the ideas immanent in nervous activity. Bulletin of Mathematical Biophysics 5, 115-133, 1943.
    5.Rosenblatt F. The perceptron: A probabilistic model for information storage and organization in the brain. Psychological Review 65(6), 386-408, 1958.
    6.B. Mehlig . Artificial neural network. arXiv:1901.05639.
    7.Rumelhart, D. E., Hinton, G. E., Williams, R. J. Learning representations by back-propagating errors. Nature 323, 533–536, 1986.
    8.Michael Nielsen. Neural Networks and Deep Learning. http://neuralnetworksanddeeplearning.com/index.html. Last visited on Dec 2018.
    9.Alex Krizhevsky, Ilya Sutskever, Geoffrey E. Hinton. ImageNet Classification with Deep Convolutional Neural Networks. In Advances in neural information processing systems, pages 1097-1105, 2012.
    10.Hu, Jie, Li Shen, and Gang Sun. "Squeeze-and-excitation networks." arXiv preprint arXiv:1709.01507 7 (2017).
    11.Park, E., et al. "ILSVRC-2017." URL http://www. image-net. org/challenges/LSVRC/2017 (2017).
    12.Waibel, Alexander, et al. "Phoneme recognition using time-delay neural networks." Readings in speech recognition. 1990. 393-404.
    13.LeCun, Yann, et al. "Gradient-based learning applied to document recognition." Proceedings of the IEEE 86.11 (1998): 2278-2324.
    14.S. Hochreiter and J. Schmidhuber. Long short-term memory. Neural computation, 9(8):1735–1780, 1997.
    15.O. Ronneberger, P. Fischer, and T. Brox, “U-Net: Convolutional networks for biomedical image segmentation,” in Proc. Int. Conf. Med.Image Comput. Comput. Assist. Intervent., Oct. 2015, pp. 234–241.
    16.Chen, K., Wang, J., Pang, J., Cao, Y., Xiong, Y., Li, X., ... & Zhang, Z. (2019). MMDetection: Open mmlab detection toolbox and benchmark. arXiv preprint arXiv:1906.07155.
    17.Ross Girshick, Ilija Radosavovic, Georgia Gkioxari, Piotr Dollar, and Kaiming He. Detectron. ´ https://github.com/facebookresearch/detectron, 2020.
    18.Francisco Massa and Ross Girshick. maskRCNN-benchmark:Fast, modular reference implementation of instance segmentation and object detection algorithms in pytorch.https://github.com/facebookresearch/maskRCNN-benchmark, 2020.
    19.Xian Tao , Dapeng Zhang, Zihao Wang , Xilong Liu , Hongyan Zhang, and De Xu. Detection of Power Line Insulator Defects Using Aerial Images Analyzed With Convolutional Neural Networks, Senior Member, IEEE 2018.
    20.S. Ren, K. He, R. Girshick, and J. Sun, “Faster R-CNN: Towards real-time object detection with region proposal networks,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 39, no. 6, pp. 1137–1149, Jun. 2017.
    21.N. Silberman, D. Hoiem, P. Kohli, and R. Fergus, “Indoor segmentation and support inference from RGBD images,” in Proc. Eur. Conf. Comput. Vis. (ECCV), Sep. 2012, pp. 746–760.
    22.J. Dai, Y. Li, K. He, and J. Sun, “R-FCN: Object detection via region-based fully convolutional networks,” in Proc. Conf. Neural Inf. Process.Syst. (NIPS), Dec. 2016, pp. 379–387.
    23.Jiaming Han, Zhong Yang, Qiuyan Zhang, Cong Chen, Hongchen Li, Shangxiang Lai, Guoxiong Hu, Changliang Xu, Hao Xu, Di Wang and Rui Chen,”A Method of Insulator Faults Detection in Aerial Images for High-Voltage Transmission Lines Inspection”,Appl, Sci, 2019, 9(10), 2009;https://doi.org/10.3390/app9102009.23
    24.Redmon, J.; Farhadi, A. Yolov3: An incremental improvement. arXiv 2018, arXiv:1804.02767.
    25.Lin, M.; Chen, Q.; Yan, S. Network in network. arXiv 2013, arXiv:1312.4400.
    26.Hao Wang, Guodong Yang , En Li , Yunong Tian, Meng Zhao , Zize Liang; ‘High-Voltage Power Transmission Tower Detection Based on Faster R-CNN and YOLO-V3’, Proceedings of the 38th Chinese Control Conference July 27-30, 2019, Guangzhou, China.
    27.Xinyu Liu, Xiren Miao, Hao Jiang, Member, IEEE, Jing Chen.’ Review of data analysis in vision inspection of power lines with an in-depth discussion of deep learning technology’, arXiv:2003.09802v1 [cs.CV] 22 Mar 2020.
    28.https://github.com/InsulatorData/InsulatorDataSet.
    29.Siyuan Qiao and Huiyu Wang and Chenxi Liu and Wei Shen and Alan Yuille, “weightstandardization”, arXiv:1903.10520, 2019.
    描述: 碩士
    國立政治大學
    資訊科學系
    107753023
    資料來源: http://thesis.lib.nccu.edu.tw/record/#G0107753023
    資料類型: thesis
    DOI: 10.6814/NCCU202101299
    顯示於類別:[資訊科學系] 學位論文

    文件中的檔案:

    檔案 描述 大小格式瀏覽次數
    302301.pdf3089KbAdobe PDF20檢視/開啟


    在政大典藏中所有的資料項目都受到原著作權保護.


    社群 sharing

    著作權政策宣告 Copyright Announcement
    1.本網站之數位內容為國立政治大學所收錄之機構典藏，無償提供學術研究與公眾教育等公益性使用，惟仍請適度，合理使用本網站之內容，以尊重著作權人之權益。商業上之利用，則請先取得著作權人之授權。
    The digital content of this website is part of National Chengchi University Institutional Repository. It provides free access to academic research and public education for non-commercial use. Please utilize it in a proper and reasonable manner and respect the rights of copyright owners. For commercial use, please obtain authorization from the copyright owner in advance.

    2.本網站之製作，已盡力防止侵害著作權人之權益，如仍發現本網站之數位內容有侵害著作權人權益情事者，請權利人通知本網站維護人員(nccur@nccu.edu.tw)，維護人員將立即採取移除該數位著作等補救措施。
    NCCU Institutional Repository is made to protect the interests of copyright owners. If you believe that any material on the website infringes copyright, please contact our staff(nccur@nccu.edu.tw). We will remove the work from the repository and investigate your claim.
    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - 回饋