臺灣博碩士論文加值系統

目錄
摘要 I
Abstract II
目次 III
圖目次 V
表目次 VI
第一章 緒論 1
1.1研究背景 1
1.2研究動機 4
1.3研究目的 5
1.4本文貢獻 6
1.5論文架構 7
第二章 文獻探討 8
2.1物件偵測簡介 8
2.2 YOLOv3突出特點 10
2.3相關文獻探討 14
第三章 研究方法與模型設計 17
3.1問題環境與訓練流程 17
3.1.1問題環境描述 17
3.1.2整體流程架構概述 20
3.2 mAP測試原理及混淆矩陣 22
3.3類別標籤與分類設計 24
3.3.1分類A之情況設計 24
3.3.2分類B之情況設計 25
3.3.1分類C之情況設計 28
3.4 加入Canny edge detection邊緣檢測算法 29
3.5模型分類 33
第四章 結果與討論 34
4.1軟硬體配備與資料集統計 34
4.1.1軟硬體設備 34
4.1.2資料集取得 34
4.2模型測試 35
4.2.1測試結果統計 35
4.2.2檢測成果圖 37
4.3 mAP成果圖 41
4.4成果討論 44
第五章 結論與未來展望 45
5.1結論 45
5.2未來展望 46
參考文獻 47
附錄一 訓練簡易流程圖 50
附錄二 Canny邊緣檢測演算法流程圖 52
附錄三 各分類PR曲線圖 54
附錄四 M4模型信心指數調整 57

圖目次
圖1.1 人工智慧示意圖 1
圖1.2 深度學習神經網路示意圖 2
圖1.3 物件偵測演算法發展史 3
圖1.6 本表數字僅含A1類[30] 5
圖2.1 RCNN慢慢達成end-to-end的學習 8
圖2.2 SSD grid cell與default box示意圖 9
圖2.3 IOU示意圖 10
圖2.4 Yolov3預測Bbox示意圖[3] 11
圖2.5 殘差網路示意圖 12
圖3.1 問題環境流程圖 17
圖3.2 環境示意圖 18
圖3.3 資料蒐集與影片分割示意圖 21
圖3.4 整體流程架構圖 22
圖3.5 Kernel=5高斯濾波器示意圖 30
圖3.6 假設求出之角度為45度，之後經過NMS的結果 31
圖3.7 邊緣像素範例圖 31
圖4.1 M4模型無車禍檢測圖[33] 37
圖4.2 M4模型無車禍檢測圖[33] 37
圖4.3 M4模型車輛翻覆檢測成果圖[33] 38
圖4.4 M3模型車輛毀損檢測成果圖[33] 38
圖4.5 M4單張清晰車禍照片測試 42

表目次
表2.1 環境視角比較 18
表3.1 資料收集比較 20
表3.2 混淆矩陣 22
表3.3 分類A標記[33] 24
表3.4 分類B設計模型B1[33] 25
表3.5 分類B設計模型B2[33] 26
表3.6 分類B設計模型B3[33] 27
表3.7 分類C標記[33] 28
表3.8 Canny過後的車輛毀損圖片 32
表3.9 Canny過後的車輛翻覆圖片 32
表3.10 Canny數據集調整閥值 33
表3.11 模型與分類設計表格 33
表4.1 硬體配置 34
表4.2 詳細資料集總數統計 34
表4.3 模型測試結果統計 36
表4.4 檢測圖方框定義 37
表4.5 特殊環境因素檢測結果[33] 40
表4.6 眾模型mAP成果圖 42
表4.7 單張清晰照片成果示意圖 43

[1] Joseph Redmon, Santosh Divvala, Ross Girshick, Ali Farhadi "You Only Look
Once:Unified,Real-Time Object Detection”IEEE International Conference on image Processing(ICIP),pp1-10,June 2016.
[2] J. Redmon and A. Farhadi, "YOLO9000: Better, Faster, Stronger," 2017 IEEE
Conference on Computer Vision and Pattern Recognition (CVPR), Honolulu, HI, 2017, pp. 6517-6525.
[3] Joseph Redmon, Ali Farhadi “Yolov3: An incremental improvement”arXiv
preprint arXiv:1804.02767, pp1-5,April 2018.
[4] Daxin Tian, Chuang Zhang, Xuting Duan, Xixian Wang “An Automatic Car Accident Detection Method Based on Cooperative Vehicle Infrastructure Systems” IEEE Journals & Magazines, pp127432-127432, September 2019.
[5] Deeksha Gour, Amit Kanskar “Optimized-YOLO:Algorithm for CPU to
Detect Road Traffic Accident and Alert System”International Journal of
Engineering Research & Technology(IJERT), Vol.8 Issue 09, September-2019.
[6] Dai W C, Jin L X, Li G N, et al. Real-time airplane detection algorithm in remote-sensing images based on improved YOLOv3[J]. Opto-Electronic
Engineering, 2018.
[7] S. Sonal and S. Suman, "A Framework for Analysis of Road Accidents," 2018 International Conference on Emerging Trends and Innovations In Engineering And Technological Research (ICETIETR), Ernakulam, 2018, pp. 1-5 .
[8] Hasan Mehdi Naqvi, Geetam Tiwari,”Factors Contributing to Motorcycle Fatal
Crashes on National Highways in India”,World Conference on Transport
Research-WCTR 2016 Shanghai, 10-15, July 2016.
[9] Wang, Chen & Yulu, Dai & Zhou, Wei & Geng, Yifei. (2020). A Vision-Based Video Crash Detection Framework for Mixed Traffic Flow Environment Considering Low-Visibility Condition. Journal of Advanced Transportation. 2020.
[10] V. Machaca Arceda and E. Laura Riveros, "Fast car Crash Detection in Video," 2018 XLIV Latin American Computer Conference (CLEI), São Paulo, Brazil, 2018, pp. 632-637.
[11] Danelljan, M., Häger, G., Khan, F., Felsberg, M.: Accurate scale estimation for robust visual tracking. In: BMVC 2014 (2014).
[12] Ravindran, Vaishnavi & Viswanathan, Lavanya & Rangaswamy, Shanta. (2016). A Novel Approach to Automatic Road-Accident Detection using Machine Vision Techniques. International Journal of Advanced Computer Science and Applications. 7. Computer Science and Applications, Vol.7, No.11, November, 2016.
[13] Fu-Hsiang Chan, Yu-Ting Chen, Ty Xiang, Min Sun, “Anticipating Accidents in Dashcam Videos”,Computer Vision-ACCV 2016:13th Asian Conference on Computer Vision, Taipei, Taiwan, November20-24, 2016.
[14] Babu, Ch. Rajesh and G. Anirudh. “Vehicle Traffic Analysis Using Yolo.”
Eurasian Journal of Analytical Chemistry 13 (2019): 345-350.
[15] J. Canny, "A Computational Approach to Edge Detection," in IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-8, no. 6, pp. 679-698, Nov. 1986 .
[16] H. Sharma, R. K. Reddy and A. Karthik, "S-CarCrash: Real-time crash detection analysis and emergency alert using smartphone," 2016 International Conference on Connected Vehicles and Expo (ICCVE), Seattle, WA, 2016, pp. 36-42.
[17] T. H. Yee and P. Y. Lau, "Mobile vehicle crash detection system," 2018
International Workshop on Advanced Image Technology (IWAIT), Chiang Mai, 2018, pp. 1-4.
[18] Pinaki Pratim Acharjya, Ritaban Das and Dibyendu Ghoshal, ―Study and Comparison of Different Edge Detectors for Image Segmentation‖, Global Journal of Computer Science and Technology Graphics & Vision, Volume 12 Issue 13, pp. 28-32, 2012.
[19] S. Singh and R. Singh, "Comparison of various edge detection techniques," 2015
2nd International Conference on Computing for Sustainable Global Development (INDIACom), New Delhi, 2015, pp. 393-396.
[20] T. Tahmid and E. Hossain, "Density based smart traffic control system using
canny edge detection algorithm for congregating traffic information," 2017 3rd International Conference on Electrical Information and Communication Technology (EICT), Khulna, 2017, pp. 1-5.
[21] C. Y. Low, H. Zamzuri and S. A. Mazlan, "Simple robust road lane detection
algorithm," 2014 5th International Conference on Intelligent and Advanced Systems (ICIAS), Kuala Lumpur, 2014, pp. 1-4.
[22] D. Tian, C. Zhang, X. Duan and X. Wang, "An Automatic Car Accident
Detection Method Based on Cooperative Vehicle Infrastructure Systems,"
in IEEE Access, vol. 7, pp. 127453-127463
[23] opencv Canny Edge Detector，參見網站
https://docs.opencv.org/2.4/doc/tutorials/imgproc/imgtrans/canny_detector/canny_detector.html#explanation


[24] 從零開始配置Yolov3(keras)訓練測試自己的數據，參見網站
https://zongxp.blog.csdn.net/article/details/86467456
[25] keras+yolo3+python+(win)訓練自己的數據，參見網站
https://blog.csdn.net/weixin_43717579/article/details/85108925
[26] WHO 2013年車禍死亡人數統計，參見網站
https://www.who.int/gho/road_safety/mortality/traffic_deaths_number/en/
[27] Artificial Intelligence,Machine Learning,and Deeplearning:Same
Context,Different concepts,參見網站
https://master-iesc-angers.com/artificial-intelligence-machine-learning-and-deep-learning-same-context-different-concepts/
[28] Gradient descent local minima,參見網站
https://bdtechtalks.com/2020/04/27/deep-learning-mode-connectivity-adversarial-attacks/gradient-descent-local-minima/
[29] Z. Zou, Z. Shi, Y. Guo and J. Ye, "Object detection in 20 years: A survey" in
arXiv:1905.05055v2, 2019.
[30] 臺灣歷年交通事故統計資料，參見網站
http://stat.motc.gov.tw/mocdb/stmain.jsp?sys=100&funid=b3303
[31] W. Liu, D. Anguelov, D. Erhan, C. Szegedy and S. Reed, "SSD: Single shot
multibox detector", 2015.
[32] Feature Pyramid Network，參見網站
https://www.twblogs.net/a/5d2334f6bd9eee1e5c83e2a4
[33] Car Crashes Time YouTube channel，參見網站
https://www.youtube.com/user/CarCrashesTime
[34] G. M. H. Amer and A. M. Abushaala, "Edge detection methods," 2015 2nd
World Symposium on Web Applications and Networking (WSWAN), Sousse,
2015, pp. 1-7.
[35] Darknet，參見網站
https://pjreddie.com/darknet/yolo/
[36] 本論文測試結果，參見網站
https://youtu.be/9-BuLzjNsHA
[37] A Person Injury Law Firm Representing Injured People,參見網站
https://www.edgarsnyder.com/car-accident/types-of-accidents/
[38] Bochkovskiy, Alexey, Chien-Yao Wang and Hong-Yuan Mark Liao. “YOLOv4:
Optimal Speed and Accuracy of Object Detection.” ArXiv abs/2004.10934 2020.