隨著台灣半導體產業之蓬勃發展，在台灣半導體產業，亦因勞動成本上漲與缺工問題，加速相關業者轉型「智慧製造工廠」方向努力，逐步透過自動化生產設備與數據資料分析…等技術，進行生產技術提升與優化作業，逐步朝「無人工廠」目標前進。研究依據影像辨識技術，完成建置第三方辨識輔助系統，透過影像深度學習方式，運用Pix2Pix Conditional GAN演算法與SSD演算法，辨識出產品PAD外框與物件針痕位置或採用人工Label標註影像方式產生模型。
優化因機台辨識作業限制之問題點，提升與改善機台影像辨識異常，生產製造端人員可降低因機台誤判，造成重複確認查檢之作業，且第三方辨識輔助系統可提供即時針痕辨識確認狀況，以釐清是否有生產與辨識異常之問題點；生產管理端更可透過辨識系統之輔助改善生產製造環境，監控針痕辨識異常率狀況與進行優化，生產製造成本上更可獲得節省及有效進行人力配置之相關效益；經改善後其AI影像辨識系統之影像辨識正確率為98.76%，透過第三方輔助影像辨識之作業方式，改善原機台自主判讀針痕異常問題點；其預估每月可增加公司整體營收約900萬元；藉由系統的開發方式，更可推展到其他不同製程之影像辨識作業，作為未來規劃參考模式之依據。

With strong development in the semiconductor industry, Taiwan is facing high labor cost and labor shortage problems. They are triggering relative companies' aim to transform into “smart manufacturing” by using high automatic production facilities, data analysis technology… to optimize production technology, targeting to realize unmanned factories to solve the problems.
The research is based on image recognition technology which implemented third party recognition assistant system. According to the image Deep Learning method, use Pix2Pix Conditional GAN and SSD algorithms to identify outer center on Pad and position of Probe Mark, Through AI image recognition system to conduct systematic secondary filter and recheck. Double confirm abnormal original images whether is「actual」abnormal, Production operations could reduce machine misjudgment caused by repeated confirmation and inspection process, To provide real time probe mark status to clarify any abnormal issues just in time Production management can improve manufacturing environment through the system and monitor the ratio of probe mark abnormal status and optimize it. After the improvement, the image recognition accuracy rate is 98.76%, the third party determine assisting system, it's optimized the limitation issue of machines, reduced abnormal determination of the machine, it is estimated that the company's overall revenue can be increased about 9 million per month. When it comes to manufacturing cost, from production cost perspective could saving and gain more effective way on manpower allocation. By system developments, it could extend to other images identify process as future reference.

摘要.....I
ABSTRACT.....II
誌謝.....III
目錄.....IV
圖目錄.....VI
表目錄.....VIII
第一章 緒論.....1
1-1 研究背景.....1
1-2 研究動機.....2
1-3 研究目的.....2
1-4 論文架構.....3
第二章 文獻回顧探討.....5
2-1 晶圓測試.....5
2-2 電腦視覺.....8
2-3 物件檢測方法.....12
2-4 文獻小結.....15
第三章 研究方法.....15
3-1 研究架構.....15
3-2 現況分析.....16
3-2-1 問題確認.....17
3-2-2 面臨挑戰.....18
3-3 問題原因分析.....18
3-3-1 特性要因分析.....18
3-3-2 動作流程分析.....19
3-3-3 異常根本原因分析.....21
3-4影像辨識.....22
3-4-1 影像辨識原理.....22
3-4-2 模型訓練作業.....28
第四章 個案分析.....33
4-1個案公司背景說明.....33
4-2 現況分析.....33
4-2-1 問題確認.....34
4-2-2 面臨挑戰.....37
4-3 問題原因分析.....39
4-3-1 特性要因分析.....39
4-3-2 動作流程分析.....40
4-3-3 異常根本原因分析.....42
4-4 影像辨識.....45
4-4-1 前置處理作業.....45
4-4-2 模型訓練作業.....46
4-4-3 模型驗證.....48
4-5 辨識系統運作架構.....49
4-5-1 EAP監控系統機制建立(Equipment Application Programming).....50
4-5-2 系統覆判異常停機警示機制建立.....50
4-5-3 針痕影像辨識系統運作架構.....51
4-5-4 辨識系統觸發停機時間運作驗證.....53
4-6 異常監控處理程序.....53
4-6-1 新Unknown Device處置.....53
4-6-2 異常監控處置.....54
4-6-3 辨識結果判讀.....57
4-6-4 異常率監控.....57
4-7辨識系統上線狀況.....58
4-7-1 系統辨識狀況.....58
4-7-2 效益評估.....59
4-8系統建置成本.....60
第五章 結論與未來研究方向.....61
5-1 結論.....61
5-2 未來研究方向.....61
參考文獻.....62
中文文獻.....62
英文文獻.....63

中文文獻
1. 丁世杰 (2012)資料探勘技術在晶圓針測誤宰分析之應用，交通大學，碩士論文
2. 廖裕評 & 陸瑞強 (2009) 積體電路測試實務，全華圖書股份有限公司
3. 邱進東 & 張達元 (2021) 垂直式晶圓探針卡之探針接觸力實驗與變形 分析，中國機械工程學刊第四十二卷第五期，第501-507頁。
4. 周敏傑 & 黃萌祺 & 高端環 & 黃悅真 & 黃鄭隆 (2021) 晶圓測試探針卡設計與製造，機械工業雜誌 459期，第16-25頁。
5. 陳世宗 & 籃山明 (2011) 探針接觸特性對於晶圓良率測試之研究，先進工程學刊6卷2期，第103-109頁。
6. 張柏司 (2017) 應用CBR模式建立維修案例庫以提升工程師維修能力-以K公司為例，交通大學，碩士論文
7. 莊弘祥 (2017) OpenCV範例解析:運用OpenCV開發實務專案 深入理解電腦視覺與影像處理，碁峯資訊股份有限公司
8. 繆鵬 (2019) CV+深度學習:AI最完整的跨套件Python人工智慧電腦視覺，深智數位股份有限公司
9. 張宸 (2020) 基於深度學習之投籃姿勢修正建議研究，淡江大學，碩士論文
10. 陳文輝 & 黃玠元 (2018) 應用於行為識別資料擴增之改良生成對抗網路，先進工程學刊 13卷3期 ( 2018/10)，第143-147頁。
11. 顏愉 & 溫欣儀 & 黃尹男 & 陳柏華 (2018) 尖端人工智慧於應用之展望，土木水利 45卷5期 ( 2018/10)，第51-58頁。
12. 洪瑞彣 (2022) 基於深度學習之自駕車挑戰賽物件辨識，淡江大學，碩士論文
13. 李宜弘 (2019) 應用迴歸式卷積神經網路於場景辨識之研究，淡江大學，碩士論文
14. 周宇軒 (2021) 基於SSD與FaceNet之人臉偵測與辨識，台北科技大學，碩士論文
15. 周明勝 (2020) 基於SSD-MobileNet交通標誌識別方法，崑山科技大學，碩士論文


英文文獻
1. Asharul Islam Khan & SalimAl-Habsi (2020) , Machine Learning in Computer Vision, Procedia Computer Science, Volume 167, 2020, P1444-1451
2. Anwaar Ulhaq & Asim Khan & Douglas Gomes & Manoranjan Paul (2020), Computer Vision For COVID-19 Control: A Survey, arXiv:2004.09420.
3. Chih Chieh Liao & Yu Min Sun & Chih Feng Cheng (2017), Probe Card, US9678110B2.
4. Cihan Ongun & Alptekin Temizel (2018), Paired 3D Model Generation with Conditional Generative Adversarial Networks, Computer Vision – ECCV 2018 Workshops, P473-487
5. David Millan Escriva & Vinicius Godoy & Prateek Joshi (2018), Learn OpenCV 4 by Building Projects: Build real-world computer vision and image processing applications with OpenCV and C++, 2nd Edition, ISBN 9781789341225.
6. Ezz El-Din Hemdan & Marwa A Shouman & Mohamed Esmail Karar (2022), Covidx-net: A framework of deep learning classifiers to diagnose covid-19 in x-ray images, arXiv:2003.11055.
7. Georgios Douzas & Fernando Bacao (2018), Effective Data Generation for Imbalanced Learning using Generative Adversarial Networks, Expert System with Application, 91, P464-471
8. Haobin Shi & Jingchen Li & Meng Liang & Maxwell Hwang & Kao-Shing Hwang & Yun-Yu Hsu (2022), Path Planning of Randomly Scattering Waypoints for Wafer Probing Based on Deep Attention Mechanism, IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2022/07/04, P1-13
9. Huan Ling & Karsten Kreis & Daiqing Li & Seung Wook Kim & Antonio Torralba & Sanja Fidler (2021), EditGAN: High-Precision Semantic Image Editing, arXiv:2111.03186.
10. Ian J. Goodfellow & Jean Pouget-Abadie & Mehdi Mirza & Bing Xu & David Warde-Farley & Sherjil Ozair & Aaron Courville & Yoshua Bengio (2014), Generative Adversarial Networks, arXiv:1406.2661.
11. Jun-Yan Zhu & Richard Zhang & Deepak Pathak & Trevor Darrell & Alexei A. Efros & Oliver Wang & Eli Shechtman (2017), Toward Multimodal Image-to-Image Translation, arXiv:1711.11586.
12. Joseph Redmon & Santosh Divvala & Ross Girshick & Ali Farhadi (2016), You Only Look Once: Unified, Real-Time Object Detetion, in Proc. of IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, P779-788
13. Lu Cheng & Yicai Ji & Chao Li & Xiaojun Liu & Guangyou Fangc (2022), Improved SSD network for fast concealed object detection and recognition in passive terahertz security images, Scientific Reports, 12, 12082 (2022).
14. Mehdi Mirza & Simon Osindero (2014), Conditional Generative Adversarial Nets, arXiv:1411.1784.
15. Noriki Koyam & Kazushi Shoji & Motokatsu Miyazaki (2018), Semiconductor system, US20180040484A1.
16. Onur Can Koyun & Reyhan Kevser Keser & İbrahim Batuhan Akkaya & Behcet Ugur Toreyin (2022), Focus-and-Detect: A small object detection framework for aerial images, Signal Processing: Image Communication, Volume 104, May 2022, arXiv:2203.12976v1.
17. Prateek Joshi (2016), OpenCV By Example, ISBN 9781785280948.
18. Phillip Isola & Jun-Yan Zhu & Tinghui Zhou & Alexei A. Efros (2017), Image-To-Image Translation With Conditional Adversarial Networks, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), P1125-1134
19. Qianhui Luo & Huifang Ma1 & Li Tang & Yue Wang & Rong Xiong (2020), 3D-SSD: Learning hierarchical features from RGB-D images for amodal 3D object detection, Neurocomputing, 378, P364-374
20. Ross Girshick & Jeff Donahue & Trevor Darrell & Jitendra Malik (2014), Rich feature hierarchies for accurate object detection and semantic segmentation, ISBN 9781479951185.
21. Ross Girshick (2015), Fast R-CNN, The IEEE International Conference on computer vision (ICCV), P1440-14486
22. Shu-Kai S. Fan & Chun-Wei Cheng & Du-Ming Tsai ( 2021), Fault Diagnosis of Wafer Acceptance Test and Chip Probing Between Front-End-of-Line and Back-End-of-Line Processes, IEEE Transactions on Automation Science and Engineering, 2021/ 08/ 31, P1-15
23. Shaoqing Ren & Kaiming He & Ross Girshick & Jian Sun (2016), Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks, Advances in Neural Information Processing System 28, Volume 1, P91-99
24. Wei Liu & Dragomir Anguelov & Dumitru Erhan & Christian Szegedy & Scott Reed & Cheng-Yang Fu & Alexander C. Berg (2016), SSD:Single Shot MultiBox Detector, arXiv:1512.02325.
25. Yifan Liu & BingHang Lu & Jingyu Peng & Zihao Zhang (2020), Research on the Use of YOLOv5 Object Detection Algorithm in Mask Wearing Recognition, World Scientific Research Journal, Volume 6 Issue 11, P 276-284