請用此 Handle URI 來引用此文件：
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/99225完整後設資料紀錄
| DC 欄位 | 值 | 語言 |
|---|---|---|
| dc.contributor.advisor | 李佳翰 | zh_TW |
| dc.contributor.advisor | Jia-Han Li | en |
| dc.contributor.author | 劉晏妤 | zh_TW |
| dc.contributor.author | Yen-Yu Liu | en |
| dc.date.accessioned | 2025-08-21T16:52:58Z | - |
| dc.date.available | 2025-08-22 | - |
| dc.date.copyright | 2025-08-21 | - |
| dc.date.issued | 2025 | - |
| dc.date.submitted | 2025-08-06 | - |
| dc.identifier.citation | [1] M. P. Jensen, C. D. Allen, T. Eguchi, I. P. Bell, E. L. LaCasella, W. A. Hilton, C. A. M. Hof, and P. H. Dutton. Environmental warming and feminization of one of the largest sea turtle populations in the world. Current Biology, 28(1):154–159.e4, 2018.
[2] Qamar A. Schuyler, Britta D. Hardesty, Chris Wilcox, and Kathy A. Townsend. A quantitative analysis linking sea turtle mortality and plastic debris ingestion. Scientific Reports, 8(1):12536, 2018. [3] B. P. Wallace, R. L. Lewison, S. L. McDonald, R. K. McDonald, C. Y. Kot, S. Kelez, R. K. Bjorkland, E. M. Finkbeiner, S. Helmbrecht, and L. B. Crowder. Global patterns of marine turtle bycatch. Conservation Letters, 3(3):131–142, 2010. [4] 海洋委員會海洋保育署. 112 年全臺海龜擱淺通報紀錄. https://www.oca.gov.tw/ch/index.jsp, 2023. 資料集來源：政府資料開放平台. [5] Karen A. Bjorndal. Foraging ecology and nutrition of sea turtles. The Biology of Sea Turtles, 1:199–231, 1997. [6] 法務部全國法規資料庫. 野生動物保育法, 2024. 民國 113 年 2 月 7 日修正. [7] Kenneth H. Pollock. Statistical inference for capture-recapture experiments. Wildlife Monographs, 107:3–97, 1990. [8] J. Reisser, M. Proietti, P. Kinas, and I. Sazima. Photographic identification of sea turtles: method description and validation, with an estimation of tag loss. Endangered Species Research, 5:73–82, 2008. [9] Annette C. Broderick and Brendan J. Godley. Effect of tagging marine turtles on nesting behaviour and reproductive success. Animal Behaviour, 58(3):587–591, 1999. [10] A. Chabrolle and E. Dumont-Dayot. Photo-identification of sea turtles in the Caribbean French West Indies. 2015. Document prepared by Higuero Emilie (Kap Natirel). [11] Felix Patton and Martin Jones. Determining the suitability of using eye wrinkle patterns for the accurate identification of individual black rhinos. Pachyderm, 48:18–23, 2010. Accessed 2024-05-17. [12] Peter Bennett and Ursula Keuper-Bennett. The Book of Honu: Enjoying and Learning About Hawaii’s Sea Turtles. Latitude 20 Books, Honolulu, Hawaii, illustrated edition, August 2008. Paperback. [13] 海龜點點名團隊. 海龜點點名 tspot：臺灣海龜攝影辨識公民科學平台, 2015. Accessed 2025-05-17. [14] Alice S. Carpentier, Claire Jean, Mathieu Barret, Agathe Chassagneux, and Stéphane Ciccione. Stability of facial scale patterns on green sea turtles Chelonia mydas over time: a validation for the use of a photo-identification method. Journal of Experimental Marine Biology and Ecology, 476:15–21, 2016. [15] Seh-Ling Long. Identification of a dead green turtle (Chelonia mydas) using photographic identification. Marine Turtle Newsletter, (150):4–6, 2016. [16] Claire Jean, Stéphane Ciccione, Etienne Talma, Kevin Ballorain, and Jerome Bourjea. Photo-identification method for green and hawksbill turtles - first results from Reunion. Indian Ocean Turtle Newsletter, 11:8–13, 2010. [17] Seh-Ling Long and Nazirul A. Azmi. Using photographic identification to monitor sea turtle populations at Perhentian Islands Marine Park in Malaysia. Herpetological Conservation and Biology, 12(2):350–366, 2017. [18] Alexey Bochkovskiy, Chien-Yao Wang, and Hong-Yuan Mark Liao. YOLOv4: Optimal speed and accuracy of object detection. arXiv preprint arXiv:2004.10934, 2020. [19] Ultralytics. Ultralytics YOLOv5 GitHub Issue 280. https://github.com/ultralytics/yolov5/issues/280, 2020. Accessed: 2025-06-04. [20] Haiying Liu, Fengqian Sun, Jason Gu, and Lixia Deng. SF-YOLOv5: A lightweight small object detection algorithm based on improved feature fusion mode. Sensors, 22(15):5817, 2022. [21] Chien-Yao Wang, Alexey Bochkovskiy, and Hong-Yuan Mark Liao. YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors. arXiv preprint arXiv:2207.02696, 2022. [22] Ultralytics. YOLOv8: Cutting-edge object detection at real-time speed. https://docs.ultralytics.com/models/yolov8/, 2023. [23] Huyen Trang Dinh and Eung-Tae Kim. A lightweight network based on YOLOv8 for improving detection performance and the speed of thermal image processing. Electronics, 14(4):783, 2025. [24] Viviane Cadenat, Adrien Durand-Petiteville, D. Billot, and Antoine Villemazet. Image-based tree detection for autonomous navigation in orchards. In Proceedings of the 2023 Latin American Robotics Symposium (LARS), Brazilian Symposium on Robotics (SBR), and Workshop on Robotics in Education (WRE), October 2023. [25] Connor Shorten and Taghi M. Khoshgoftaar. A survey on image data augmentation for deep learning. Journal of Big Data, 6(1):60, 2019. [26] Syed M. Abbas and Mohd Nasir Uddin. Variational image denoising approach with diffusion porous media flow. Abstract and Applied Analysis, 2013:8 pages, 2013. [27] Sangdoo Yun, Dongyoon Han, Seong Joon Oh, Sanghyuk Chun, Junsuk Choe, and Youngjoon Yoo. CutMix: Regularization strategy to train strong classifiers with localizable features. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 6023–6032, 2019. | - |
| dc.identifier.uri | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/99225 | - |
| dc.description.abstract | 本研究針對臺灣海域海龜個體辨識問題，提出一套資料增強輔助之三層級分類系統，以逐層分類方式協助縮小比對範圍、加速辨識流程。研究採用 YOLOv5、YOLOv7 與 YOLOv8 三種深度學習架構，分別訓練於物種分類（ClassA）、左眼下鱗片數量分類（ClassB）及臉部特徵分類（ClassC）等任務，並進行模型效能比較。實驗結果顯示，YOLOv8 於複雜特徵辨識任務（ClassC）中表現最佳，具備較佳的泛化能力與辨識穩定性。本研究所提出之系統具備良好之可擴充性及實務應用潛力，為未來自動化辨識系統建置提供實證基礎。 | zh_TW |
| dc.description.abstract | This study addresses the individual identification of sea turtles in Taiwan by proposing a three-tier classification system with data augmentation to assist in narrowing down candidate lists and accelerating the identification processes. The research employs three YOLO-based deep learning architectures—YOLOv5, YOLOv7, and YOLOv8—trained separately on species classification (ClassA), scute number under left eye classification (ClassB), and left facial scute pattern (ClassC) tasks, respectively. Model performance comparisons reveal that YOLOv8 achieves superior results in complex feature recognition tasks (ClassC), demonstrating better generalization and classification stability. Overall, the proposed system exhibits excellent scalability and practical potential, providing an empirical foundation for future automated identification systems. | en |
| dc.description.provenance | Submitted by admin ntu (admin@lib.ntu.edu.tw) on 2025-08-21T16:52:58Z No. of bitstreams: 0 | en |
| dc.description.provenance | Made available in DSpace on 2025-08-21T16:52:58Z (GMT). No. of bitstreams: 0 | en |
| dc.description.tableofcontents | 致謝 ............................................................. i
摘要 ............................................................. ii Abstract ......................................................... iii 研究貢獻 ....................................................... iv 目次 ............................................................. vi 圖次 ............................................................. x 表次 ............................................................. xii 第一章 緒論 ..................................................... 1 1.1 研究背景 .................................................. 1 1.1.1 全球海龜族群現況與保育挑戰 ...................... 1 1.1.2 臺灣地區海龜致死原因分析 ........................ 2 1.1.3 海龜的生態功能與臺灣保育政策 .................... 4 1.1.4 傳統族群估算方法與其侷限 ........................ 4 1.2 研究動機 .................................................. 5 1.2.1 傳統標記法之限制 ................................ 5 1.2.2 個體辨識技術之轉變：攝影辨識法 ................. 7 1.3 研究目的 .................................................. 7 1.4 論文架構簡述 .............................................. 8 第二章 文獻探討 ................................................ 10 2.1 海龜個體辨識技術與臉部特徵應用 ....................... 10 2.1.1 非侵入式的辨識技術：Photo-identification ..... 10 2.1.2 臉部鱗片作為個體辨識特徵的可行性 .............. 11 2.1.3 海龜臉部辨識流程中的分類依據 .................. 11 2.2 物件偵測與分類模型：YOLO 系列架構 .................. 13 2.2.1 YOLOv5：模型架構與技術特性 .................. 14 2.2.2 YOLOv7：模型架構與技術特性 .................. 16 2.2.3 YOLOv8：模型架構與技術特性 .................. 18 2.3 資料增強技術 ............................................ 20 2.3.1 高斯雜訊（Gaussian Noise） ................... 20 2.3.2 CutMix ............................................ 21 第三章 研究方法設計 ............................................ 23 3.1 整體流程概述 ............................................ 23 3.2 資料來源與前處理 ........................................ 24 3.2.1 資料蒐集方式 .................................... 24 3.2.2 資料篩選 ........................................ 25 3.2.3 資料分類標準 .................................... 26 3.2.4 影像前處理 .................................... 28 3.2.5 標註工具與格式 ................................ 29 3.3 三層分類架構 ............................................ 30 3.4 模型比較與評估方式 ...................................... 30 3.5 誤判定義與統計方式 ...................................... 31 3.6 系統整合與應用設計 ...................................... 32 第四章 模擬結果與分析 .......................................... 33 4.1 實作流程與設定 .......................................... 33 4.1.1 開發環境與訓練條件 ............................ 33 4.1.2 資料切分與類別分佈 ............................ 34 4.1.3 模型架構與比較設計 ............................ 34 4.2 三層分類結果分析 ........................................ 35 4.2.1 ClassA：物種分類 .............................. 35 4.2.2 ClassB：左眼下鱗片數分類 .................... 38 4.2.3 ClassC：臉部特徵分類 ........................ 41 4.2.4 資料增強：CutMix 於 YOLOv8 的應用成效 ..... 44 4.2.5 進一步模擬：資料增強組合下之 CutMix 成效評估 ........................................ 46 4.3 敏感度分析 ............................................. 48 4.3.1 Epoch 與 Batch size 之參數調整模擬 .......... 48 4.3.2 訓練參數敏感度分析 exp1（epoch=300，batch size=18） ....................... 50 4.3.3 訓練參數敏感度分析 exp2（epoch=200，batch size=18） ....................... 50 4.3.4 訓練參數敏感度分析 exp3（epoch=400，batch size=18） ....................... 51 4.3.5 訓練參數敏感度分析 exp4（epoch=300，batch size=12） ....................... 53 4.3.6 訓練參數敏感度分析 exp5（epoch=300，batch size=24） ....................... 54 4.3.7 訓練參數敏感度分析 exp6（epoch=300，batch size=32） ....................... 55 4.3.8 Loss curve 比較 ............................... 56 4.3.9 F1-score curve 比較 .......................... 57 第五章 結論與未來展望 .......................................... 59 5.1 結論 .................................................... 59 5.2 未來展望 ................................................ 60 參考文獻 ......................................................... 61 附錄 A — 模型參數表 ............................................. 65 A.1 ClassA（YOLOv5）模型設定 ......................... 65 A.2 ClassB（YOLOv7）模型設定 ......................... 65 A.3 ClassC（YOLOv8）模型設定 ......................... 66 | - |
| dc.language.iso | zh_TW | - |
| dc.subject | 海龜 | zh_TW |
| dc.subject | 個體辨識 | zh_TW |
| dc.subject | YOLO | zh_TW |
| dc.subject | 資料增強 | zh_TW |
| dc.subject | 敏感度分析 | zh_TW |
| dc.subject | 公民科學 | zh_TW |
| dc.subject | Data augmentation | en |
| dc.subject | YOLO | en |
| dc.subject | Sensitivity analysis | en |
| dc.subject | Sea turtle | en |
| dc.subject | Citizen science | en |
| dc.subject | Individual identificatio | en |
| dc.title | 針對小資料集之資料分類器增強輔助設計:以海龜為例 | zh_TW |
| dc.title | Classification Framework for Small Datasets with Data Augmentation Support: Sea Turtle as an Example | en |
| dc.type | Thesis | - |
| dc.date.schoolyear | 113-2 | - |
| dc.description.degree | 碩士 | - |
| dc.contributor.oralexamcommittee | 薛承輝;林俊宏;張恆華;謝議霆 | zh_TW |
| dc.contributor.oralexamcommittee | Chun-Hway Hsueh;Chun-Hung Lin;Herng-Hua Chang;I-Ting Hsieh | en |
| dc.subject.keyword | 海龜,個體辨識,YOLO,資料增強,敏感度分析,公民科學, | zh_TW |
| dc.subject.keyword | Sea turtle,Individual identificatio,YOLO,Data augmentation,Sensitivity analysis,Citizen science, | en |
| dc.relation.page | 66 | - |
| dc.identifier.doi | 10.6342/NTU202502906 | - |
| dc.rights.note | 同意授權(全球公開) | - |
| dc.date.accepted | 2025-08-10 | - |
| dc.contributor.author-college | 工學院 | - |
| dc.contributor.author-dept | 工程科學及海洋工程學系 | - |
| dc.date.embargo-lift | 2025-08-22 | - |
| 顯示於系所單位： | 工程科學及海洋工程學系 | |
文件中的檔案：
| 檔案 | 大小 | 格式 | |
|---|---|---|---|
| ntu-113-2.pdf | 8.91 MB | Adobe PDF | 檢視/開啟 |
系統中的文件，除了特別指名其著作權條款之外，均受到著作權保護，並且保留所有的權利。