臺灣博碩士論文加值系統

本研究藉由深度學習技術及影像辨識結合，設計一套對剎車來令片檢測判斷輪廓與位置系統。剎車來令片製造過程中因製程關係導致會有膨脹、缺角…等問題必須先找出其輪廓與位置，本研究過程先將剎車來令片分為輪廓並拍照建立圖檔，接著進行影像處理去除主體以外部分，最後將處理過的圖檔做標記並輸入至神經網路中訓練，透過損失函數觀察模型訓練狀況。本論文提出的方法可以有效達到透過影像處理解決背景雜亂多干擾問題、判別剎車來令片輪廓、自動矯正影像之目的。

In this thesis, deep learning image recognition with Tensorflow and image processing (OpenCV) are used to calculate contours and positions of brake pads for automatic detection system. In the process of brake pad manufacture, it may result in some problems, such as expansion, missing corners. Primary step is to determine contours and positions of brake pads by taking their picture to establish databases. Then, image recognition is used by using deep learning to train the contours and positions of the brake pads. Finally, the pre-processed images are labeled and input into the neural network for training. The images are evaluated through loss function in order to observe the training process of the models. The proposed method in this thesis can effectively achieve the following objectives: Solve the problem of the background multiple obstructions through image recognition; determine the brake pad contours and positions; automatically correct the image.

摘要......................................................i
Abstract..................................................ii
誌謝......................................................iii
目錄......................................................iv
圖目錄.................................................... vi
第一章 緒論..............................................1
1.1 研究動機與目的..........................................1
1.2 文獻探討...............................................3
1.3 論文大綱...............................................4
第二章 深度學習與影像辨識技術介紹...........................5
2.1 類神經網路(Neural Network , NN).........................5
2.2 卷積神經網路(Convolutional Neural Network, CNN).........9
2.3 反傳遞演算法(Back propagation)..........................12
2.4 影像處理(Image Recognition).............................14
第三章 研究架構與實驗方法...................................19
3.1 研究架構................................................19
3.2 資料蒐集................................................20
3.3 影像預處理..............................................21
3.4 尋找邊緣................................................25
3.5 資料預處理...............................................28
3.6 建立模型................................................30
第四章 實驗與結果探討.......................................35
4.1 尋找輪廓與模糊化之關係實驗結果.............................35
4.2 尋找來令片邊緣區域實驗結果.................................37
4.3 手動標記尋找邊緣實驗結果...................................38
4.4 模型實驗結果..............................................41
第五章 結論與未來研究方向.....................................46
5.1 結論......................................................46
5.2 未來研究方向...............................................47
參考文獻.......................................................48
Extended Abstract............................................50

[1]B. R. Mete and T. Ensari, "Flower Classification with Deep CNN and Machine Learning Algorithms," 2019 3rd International Symposium on Multidisciplinary Studies and Innovative Technologies (ISMSIT), Ankara, Turkey, 2019, pp. 1-5, doi: 10.1109/ISMSIT.2019.8932908.
[2]F. Ertam and G. Aydın, "Data classification with deep learning using Tensorflow," 2017 International Conference on Computer Science and Engineering (UBMK), Antalya, 2017, pp. 755-758, doi: 10.1109/UBMK.2017.8093521.
[3]Y. LeCun, K. Kavukcuoglu and C. Farabet, "Convolutional networks and applications in vision," Proceedings of 2010 IEEE International Symposium on Circuits and Systems, Paris, 2010, pp. 253-256, doi: 10.1109/ISCAS.2010.5537907.
[4]H. Shin et al., "Deep Convolutional Neural Networks for Computer-Aided Detection: CNN Architectures, Dataset Characteristics and Transfer Learning," in IEEE Transactions on Medical Imaging, vol. 35, no. 5, pp. 1285-1298, May 2016, doi: 10.1109/TMI.2016.2528162.
[5]H. Qassim, A. Verma and D. Feinzimer, "Compressed residual-VGG16 CNN model for big data places image recognition," 2018 IEEE 8th Annual Computing and Communication Workshop and Conference (CCWC), Las Vegas, NV, 2018, pp. 169-175, doi: 10.1109/CCWC.2018.8301729.
[6]林大貴，「Tensorflow+keras深度學習人工智慧實務應用」，博碩文化股份有限公司，2017年5月。
[7]黃鴻波，「Tensorflow進階指南:基礎、算法與應用」，電子工業出版社，2018年11月。
[8]Z. Zeng, Q. Gong and J. Zhang, "CNN Model Design of Gesture Recognition Based on Tensorflow Framework," 2019 IEEE 3rd Information Technology, Networking, Electronic and Automation Control Conference (ITNEC), Chengdu, China, 2019, pp. 1062-1067, doi: 10.1109/ITNEC.2019.8729185.
[9]李立宗，「OpenCV輕鬆入門:=面向Python」，電子工業出版社，2019年5月。
[10]Y. M. Wang, Y. Li and J. B. Zheng, "A camera calibration technique based on OpenCV," The 3rd International Conference on Information Sciences and Interaction Sciences, Chengdu, 2010, pp. 403-406, doi: 10.1109/ICICIS.2010.5534797.
[11]B.N.K. Sai and T. Sasikala, "Object Detection and Count of Objects in Image using Tensor Flow Object Detection API," 2019 International Conference on Smart Systems and Inventive Technology (ICSSIT), Tirunelveli, India, 2019, pp. 542-546, doi: 10.1109/ICSSIT46314.201
9.8987942.
[12]C. Szegedy, V. Vanhoucke, S. Ioffe, J. Shlens and Z. Wojna, "Rethinking the Inception Architecture for Computer Vision," 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, 2016, pp. 2818-2826, doi: 10.1109/CVPR.20
16.308.
[13]M. Sandler, A. Howard, M. Zhu, A. Zhmoginov and L. Chen, "MobileNetV2: Inverted Residuals and Linear Bottlenecks," 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, 2018, pp. 4510-4520, doi: 10.1109/CVPR.2018.00
474.