臺灣博碩士論文加值系統

生態環境問題是當今世界所面臨的一大挑戰，尤其是在過去的數十年中，由於工業化、城市化和經濟發展等因素的影響，生態環境受到了嚴重的破壞。這對生物群落和生態系統產生了深遠的影響，許多物種面臨生存威脅，生態平衡受到了破壞。因此，保護和改善生態環境已經成為全球迫切需要解決的重要問題之一。
倒立水母（Cassiopea Andromeda）屬於刺胞動物門，通常生活在潮間帶沙地、泥灘或紅樹林等區域。近年來，這些水母在高雄林園溼地公園和林園汕尾廢棄魚塭被發現，引起了人們的關注。生態環境的改變對物種數量產生了直接的影響，而倒立水母作為觀測對象，因其對乾淨水源和水域流速緩慢的需求。
本論文以野外倒立水母作為實驗對象，探討影像處理與辨識技術在生態環境監測中的應用。研究利用影像處理技術減少外在環境的干擾，然後應用深度學習演算法對水母照片進行模型訓練。實驗結果顯示，相比於現有論文的辨識率，本論文的方法能夠更精確地辨識水母。
這項研究不僅擴展了我們對倒立水母生態的認識，也為生態環境監測提供了一種新的技術手段。透過先進的影像處理和深度學習技術，我們能夠更有效地監測和保護這些生態系統中的物種。這不僅有助於維護生態平衡，還能為環境保護和永續發展提供寶貴的數據和方法。

Environmental issues are a major challenge facing the world today, especially in the past few decades. The impact of factors such as industrialization, urbanization, and economic development has led to severe environmental degradation. This has had profound effects on biological communities and ecosystems, posing a threat to the survival of many species and disrupting ecological balance. Therefore, the protection and improvement of the environment have become urgent and critical global issues.
The upside-down jellyfish (Cassiopea Andromeda) belongs to the phylum Cnidaria and is commonly found in intertidal sandy areas, mudflats, and mangrove regions. In recent years, these jellyfish have been discovered in the wetland park in Lin Yuan, Kaohsiung, and abandoned fish ponds in Lin Yuan Shanwei, attracting attention. Changes in the ecological environment directly impact species abundance, and the upside-down jellyfish is observed due to its specific requirements for clean water sources and slow water flow.
This paper focuses on the application of image processing and recognition technology in ecological monitoring, using wild upside-down jellyfish as the experimental subject. The study utilizes image processing techniques to reduce external environmental interference, followed by applying deep learning algorithms to train models on jellyfish images. Experimental results demonstrate that, compared to existing literature, this paper's method can more accurately identify jellyfish.
This research not only expands our understanding of the ecology of upside-down jellyfish but also provides a new technological approach for ecological environmental monitoring. Through advanced image processing and deep learning techniques, we can more effectively monitor and protect species in these ecosystems. This not only contributes to maintaining ecological balance but also provides valuable data and methods for environmental conservation and sustainable development.

摘要.....i
Abstract.....ii
誌謝.....iv
目錄.....v
表目錄.....vii
圖目錄.....viii
符號說明.....x
第一章 緒論.....1
1.1 生態環境概論.....1
1.1.1 生態系統.....1
1.1.2 生態平衡.....2
1.2 影像處理概論.....2
1.2.1 影像識別.....2
1.2.2 影像處理之應用.....3
1.3 深度學習概論.....3
1.3.1 深度學習框架.....4
1.3.2 神經網路（Neural Networks）.....4
1.4 研究目的與動機.....5
1.5 論文架構.....5
第二章 相關背景知識.....6
2.1 相關文獻.....6
2.2 相關背景知識.....7
2.2.1 Faster R-CNN(Region-based Convolutional Neural Network).....8
2.2.2 激勵函數（Activation Function）.....9
第三章 適用於野外倒立水母檢測方式.....10
3.1 影像反射消除處理.....11
3.2 適用於倒立水母的YOLO訓練方法.....18
3.3 物件檢測.....21
第四章 實驗結果與比較.....24
4.1 實驗設備及處理平台.....24
4.2 實驗數據.....25
4.3 實驗結果分析比較.....26
4.3.1 使用Faster R-CNN的分析比較.....26
4.3.2 使用Single Image Reflection Removal的分析比較.....36
4.3.3 對激勵函數的調整及分析.....45
第五章 結論與未來展望.....68
參考文獻.....69
Extended Abstract.....72
Chapter 1 Introduction.....74
Chapter 2 Related background knowledge.....74
Chapter 3 YOLO Training Method for Inverted Jellyfish.....75
Chapter 4 Experimental Results and Discussion.....75
Chapter 5 Conclusion and Future Work.....76

[ 1 ]Francisco E. Fontúrbel, Roberto F. Nespolo, Guillermo C. Amico, David M. Watson, "Climate change can disrupt ecological interactions in mysterious ways: Using ecological generalists to forecast community-wide effects." Climate Change Ecology, vol. 2, 2021, p. 100044. ISSN 2666-9005.
[ 2 ]Ömer Esen, Durmuş Çağrı Yıldırım, and Seda Yıldırım, "Pollute less or tax more? Asymmetries in the EU environmental taxes – Ecological balance nexus." Environmental Impact Assessment Review, vol. 91, 2021, p. 106662. ISSN 0195-9255.
[ 3 ] Zonglin Yang, Yaqi Shi, Jun Wang, Le Wang, Xianguo Li, Dahai Zhang, "Unique functional responses of fungal communities to various environments in the mangroves of the Maowei Sea in Guangxi, China." Marine Pollution Bulletin, vol. 173, part B, 2021, p. 113091. ISSN 0025-326X.
[ 4 ]Nguyen, Anh et al. “Long-Term Heavy Metal Retention by Mangroves and Effect on Its Growth: A Field Inventory and Scenario Simulation.” International journal of environmental research and public health vol. 17,23 9131. 7 Dec. 2020, doi:10.3390/ijerph17239131
[ 5 ]Juan Carlos Vizcarra, Erik A. Burlingame, Clemens B. Hug, Yury Goltsev, Brian S. White, Darren R. Tyson, Artem Sokolov, "A community-based approach to image analysis of cells, tissues and tumors." Computerized Medical Imaging and Graphics, vol. 95, 2022, p. 102013. ISSN 0895-6111.
[ 6 ]Bochkovskiy, Alexey, Chien-Yao Wang, and Hong-Yuan Mark Liao. "YOLOv4: Optimal Speed and Accuracy of Object Detection." arXiv preprint arXiv:2004.10934 (2020).
[ 7 ]Ren, Shaoqing, et al. "Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks." IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), vol. 39, no. 6, 2017, pp. 1137-1149.
[ 8 ]Mrinal R. Bachute, Javed M. Subhedar, "Autonomous Driving Architectures: Insights of Machine Learning and Deep Learning Algorithms." Machine Learning with Applications, vol. 6, 2021, p. 100164. ISSN 2666-8270.
[ 9 ]Xue Lv, Mingxia Su, Zekun Wang, "Application of Face Recognition Method Under Deep Learning Algorithm in Embedded Systems." Microprocessors and Microsystems, 2021, p. 104034. ISSN 0141-9331.
[ 10 ]Hsieh-Chang Huang, Ching-Tang Hsieh, Min-Nan Hsiao, Cheng-Hsiang Yeh, "A Study of Automatic Separation and Recognition for Overlapped Fingerprints." Applied Soft Computing, vol. 71, 2018, pp. 127-140. ISSN 1568-4946.
[ 11 ]Yifeng Chen, Cheng Wu, Yiming Wang, "Whether Normalized or Not? Towards More Robust Iris Recognition Using Dynamic Programming." Image and Vision Computing, vol. 107, 2021, 104112. ISSN 0262-8856.
[ 12 ]Martín Abadi, Paul Barham, Jianmin Chen, Zhifeng Chen, Andy Davis, Jeffrey Dean, Matthieu Devin, Sanjay Ghemawat, Geoffrey Irving, Michael Isard, Manjunath Kudlur, Josh Levenberg, Rajat Monga, Sherry Moore, Derek G. Murray, Benoit Steiner, Paul Tucker, Vijay Vasudevan, Pete Warden, Martin Wicke, Yuan Yu, and Xiaoqiang Zheng. "TensorFlow: A System for Large-Scale Machine Learning." Proceedings of the 12th USENIX Conference on Operating Systems Design and Implementation (OSDI'16). USENIX Association, 2016, pp. 265-283.
[ 13 ]Adam Paszke, Sam Gross, Francisco Massa, Adam Lerer, James Bradbury, Gregory Chanan, Trevor Killeen, Zeming Lin, Natalia Gimelshein, Luca Antiga, Alban Desmaison, Andreas Köpf, Edward Yang, Zach DeVito, Martin Raison, Alykhan Tejani, Sasank Chilamkurthy, Benoit Steiner, Lu Fang, Junjie Bai, and Soumith Chintala. "PyTorch: An Imperative Style, High-Performance Deep Learning Library." Proceedings of the 33rd International Conference on Neural Information Processing Systems. Curran Associates Inc., 2019.
[ 14 ]Chollet, François, and others. "Keras: The Python Deep Learning library." Astrophysics Source Code Library, 2018, ascl:1806.022.
[ 15 ]Yangqing Jia, Evan Shelhamer, Jeff Donahue, Sergey Karayev, Jonathan Long, Ross Girshick, Sergio Guadarrama, Trevor Darrell. "Caffe: Convolutional Architecture for Fast Feature Embedding." MM 2014 - Proceedings of the 2014 ACM Conference on Multimedia, 2014, doi:10.1145/2647868.2654889
[ 16 ]Jürgen Schmidhuber. "Deep Learning in Neural Networks: An Overview." Neural Networks, vol. 61, 2015, pp. 85-117. ISSN 0893-6080.
[ 17 ]Adam Gauci, et al. "Automating Jellyfish Species Recognition through Faster Region-Based Convolution Neural Networks." Applied Sciences, 2020, doi:10.3390/app10228257.
[ 18 ]Martin-Abadal M, Ruiz-Frau A, Hinz H, Gonzalez-Cid Y. Jellytoring: Real-Time Jellyfish Monitoring Based on Deep Learning Object Detection. Sensors. 2020; 20(6):1708.
[ 19 ]Ben Mcilwaine, et al. "JellyNet: The Convolutional Neural Network Jellyfish Bloom Detector." International Journal of Applied Earth Observation and Geoinformation, vol. 97, 2021, p. 102279.
[ 20 ]鄭鈞維。「台灣高雄林園濕地倒立水母族群之季節動態及其與水文環境關聯之研究」。碩士論文，國立中山大學海洋生物科技暨資源學系研究所，2020。
[ 21 ]Glorot, Xavier & Bordes, Antoine & Bengio, Y.. (2010). Deep Sparse Rectifier Neural Networks. Journal of Machine Learning Research. 15.
[ 22 ]Z. Dong, K. Xu, Y. Yang, H. Bao, W. Xu and R. W. H. Lau, "Location-aware Single Image Reflection Removal," 2021 IEEE/CVF International Conference on Computer Vision (ICCV), Montreal, QC, Canada, 2021, pp. 4997-5006, doi: 10.1109/ICCV48922.2021.00497.
[ 23 ]Bochkovskiy, Alexey, Chien-Yao Wang, and Hong-Yuan Mark Liao. "YOLOv4: Optimal Speed and Accuracy of Object Detection." arXiv preprint arXiv:2004.10934 (2020).
[ 24 ]Diganta Misra."Mish: A Self Regularized Non-Monotonic Activation Function." 2020. arXiv preprint arXiv:1908.08681.
[ 25 ]Maas, Andrew L.. “Rectifier Nonlinearities Improve Neural Network Acoustic Models.” (2013).
[ 26 ]Clevert, Djork-Arné & Unterthiner, Thomas & Hochreiter, Sepp. (2016). Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs).
[ 27 ]Nair, Vinod & Hinton, Geoffrey. (2010). Rectified Linear Units Improve Restricted Boltzmann Machines Vinod Nair. Proceedings of ICML. 27. 807-814.
[ 28 ]Gulcehre, Caglar & Moczulski, Marcin & Denil, Misha & Bengio, Y.. (2016). Noisy Activation Functions.
[ 29 ]Everingham, M., Van Gool, L., Williams, C.K.I. et al. The PASCAL Visual Object Classes (VOC) Challenge. Int J Comput Vis 88, 303–338 (2010). https://doi.org/10.1007/s11263-009-0275-4