在現今社會中人們的環保意識日益增強，然而，瓶裝海洋垃圾 (BMD) 仍然是世界上備受重視的環境問題之一。 傳統海灘垃圾研究中的監測方法因為人力資源的關係存在著許多局限性，因此本研究提出了一種利用無人機和物件辨識BMD的方法，在桃園市大園區的沙灘上進行了相關研究。
首先，本研究設計了三個實驗區域進行圖像收集用於模型訓練，且為了確保此方法在長期研究上的可行性，另外收集了兩處真實區域(非實驗區域)之圖像用於驗證模型強健性。接著，使用無人機於不同高度收集圖像，其解析度為 0.12 至 1.54 厘米/像素；物件辨識系統則採用You Only Look Once version 2 (YOLO v2) ，其使用無人機收集之圖像進行訓練 BMD辨識模型；此外本研究應用背景移除之影像處理技術來移除圖像中的雜訊、於訓練過程中應用數據增強(Data augmentation)之技術增加訓練數據量以提升模型可信性，並採用聯合交集（IoU）來評估訓練效率。
本研究發現在航測上使用 0.5 厘米/像素的解析度能得到最佳的結果，該解析度於實驗區域之準確率(precision)達到 0.94及召回率(recall rate)達到0.97 ，可得 0.95 的 F1-score；在真實區域上，檢測的平均準確率為 0.61，召回率為 0.86，F1-score為 0.72。 本研究顯示，數據增強之應用在訓練過程中起著至關重要的作用，其結果IoU 超過 0.68；而背景移除技術則大量節省整個檢測時間，也因為移除了大量雜訊，減少了於真實區域中檢測錯誤的情況，證實數據增強及背景移除技術可以更準確、快速和客觀地識別海灘上的垃圾。

Humans’ awareness of the environment is increasing nowadays; however, bottle marine debris (BMD) remains one of the most pressing global issues. Fields surveys of marine debris based on manpower is less efficient; therefore, this study proposes an automatic detection method on BMD using unmanned aerial vehicles and machine learning techniques.
The study sites are located on sandy beaches in Dayuan District, Taoyuan City. We first set three designed sites to create training datasets and test the detecting algorithm, and performances. Two real sites were then surveyed to evaluate our method in such a sandy complex beach that was intended to be used for long-term researches. The UAVs were operated at different fly heights to capture images with resolutions from 0.12 to 1.54 cm/pixel. The object detection algorithm You Only Look Once version 2 (YOLO v2) was trained to identify BMD and we added an image processing skill to remove image background noises. Data augmentation was used in training process to increase training data, and intersection over union (IoU) was adopted to evaluate the training efficiency. The results reveal that the skill of data augmentation helps IoU reaches over 0.68; and the skill of background removal has an advantage to reduce the processing time, as well as reducing noise resulting in much greater precision in real sites. From testing on both the designed and real sites with different image resolutions and processing skills, we found that approximately 0.5 cm/pixel could be the optimal resolution for aerial surveys on BMD. When operating the UAV with an image resolution of 0.5 cm/pixel, the performance indexes of mean precision, recall rate, and F1-score are respectively, 0.94, 0.97 and 0.95 at designed sites and are 0.61, 0.86, and 0.72 at real sites.
Our work contributes to advances in beach debris surveys, optimizes the automatic detection on machine learning approach, especially with the role of data augmentation step in training data and background removing procedure.

English Abstract I
摘要II
Acknowledgment . III
Table of Contents V
List of FiguresVII
List of Tables X
List of Acronyms and Abbreviations . XI
Chapter 1 Introduction . 1
1.1 Background. 1
1.2 Literature review 3
1.2.1 Visual census 4
1.2.2 UAV aerial surveys combined with the use of AI. 5
1.3 Motivation and objectives 15
1.4 Thesis organization 16
Chapter 2 Methodology 17
2.1 Field experiments 17
2.1.1 Study areas. 17
2.1.2 Experimental setup. 18
2.1.3 Aerial survey . 21
2.2 Process of training image . 25
2.2.1 Raw image accumulation 25
2.2.2 Image segmentation and anchor design 25
2.2.3 Machine learning for BMD detection 27
2.3 Background removal 32
2.4 Object detecting process 34
2.5 Evaluation of the detecting performance 36
2.5.1 Intersection over Union (IoU) 36
2.5.2 Precision 38
2.5.3 Recall 39
2.5.4 F1-score 40
2.5.5 Accuracy . 41
Chapter 3 Results of the designed experimental sites 42
3.1 Performance of augmentation phase 42
3.2 Effects of resolution on the performance of detecting 47
3.3 Evaluation of different designed sites. 51
3.3.1 Designed site 2 (25.0792°N 21.1524°E). 51
3.3.2 Designed site 3 (25.0791°N 121.1509°E) 54
3.4 Discussions 56
3.4.1 Effects of image types on training and detecting processes . 56
3.4.2 Effects of image resolutions and study regions’ landscape 57
Chapter 4 Applications on real field conditions 59
4.1 Real site 1 (25.0742°N 121.1291°E) 59
4.2 Real site 2 (25.0741°N 121.1292°E) 61
4.3 Discussion. 64
4.3.1 Effects of resolution on the performance of detecting. 64
4.3.2 Effects of image types on the performance of detecting 65
4.3.3 Possibility 66
Chapter 5 Conclusion and recommendation for future research . 68
5.1 Conclusion 68
5.2 Limitations 69
5.3 Recommendations for future study 70
References 72
Vita 77

References
1. (ICC), I. C. C. (2001). 2000 International Coastal Cleanup International Results. Retrieved from The Ocean Conservancy: https://oceanconservancy.org/wp-content/uploads/2017/04/2000-Ocean-Conservancy-International-Report.pdf
2. (ICC), I. C. C. (2020). Final 2019 ICC Report. Ocean Conservancy. Retrieved from https://oceanconservancy.org/wp-content/uploads/2019/09/Final-2019-ICC-Report.pdf
3. Andrady, A. L. (2011). Microplastics in the marine environment. Mar Pollut Bull, 62(8), 1596-1605. doi:10.1016/j.marpolbul.2011.05.030
4. Bao, Z., Sha, J., Li, X., Hanchiso, T., & Shifaw, E. (2018). Monitoring of beach litter by automatic interpretation of unmanned aerial vehicle images using the segmentation threshold method. Marine Pollution Bulletin, 137, 388-398.
5. Bouchard, M., Jousselme, A.-L., & Doré, P.-E. (2013). A proof for the positive definiteness of the Jaccard index matrix. International Journal of Approximate Reasoning, 54(5), 615-626.
6. Cai, Z., & Vasconcelos, N. (2019). Cascade r-cnn: High quality object detection and instance segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence.
7. Chang, Y.-L., Anagaw, A., Chang, L., Wang, Y. C., Hsiao, C.-Y., & Lee, W.-H. (2019). Ship detection based on YOLOv2 for SAR imagery. Remote Sensing, 11(7), 786.
8. Chen, H., Wang, S., Guo, H., Lin, H., Zhang, Y., Long, Z., & Huang, H. (2019). Study of marine debris around a tourist city in East China: Implication for waste management. Sci Total Environ, 676, 278-289. doi:10.1016/j.scitotenv.2019.04.335
9. Cheshire, A., Adler, E., Barbière, J., Cohen, Y., Evans, S., Jarayabhand, S., . . . Westphalen, G. (2009). UNEP/IOC Guidelines on Survey and Monitoring of Marine Litter Regional Seas Reports and Studies No. 186 IOC Technical Series No. 83. UNEP Regional Seas Reports and Studies.
10. Chinchor, N., & Sundheim, B. M. (1993). MUC-5 evaluation metrics. Paper presented at the Fifth Message Understanding Conference (MUC-5): Proceedings of a Conference Held in Baltimore, Maryland, August 25-27, 1993.
11. Chollet, F. (2018). Deep learning with Python (Vol. 361): Manning New York.
12. Cioppa, A., Braham, M., & Van Droogenbroeck, M. (2020). Asynchronous Semantic Background Subtraction. Journal of Imaging, 6(6), 50.
13. Conservancy, O. (2010). Trash travels. From our hands to the sea, around the globe, and through time. International Coastal Cleanup, 15.
14. Conservancy, O. (2016). 30th Anniversary International Coastal Cleanup.
15. Council, N. R. (1975). Assessing Potential Ocean Pollutants: A Report of the Study Panel on Assessing Potential Ocean Pollutants to the Ocean Affairs Board, Commission on Natural Resources, National Research Council: The National Academy of Science.
16. Dave, A., Khurana, T., Tokmakov, P., Schmid, C., & Ramanan, D. (2020). Tao: A large-scale benchmark for tracking any object. Paper presented at the European conference on computer vision.
17. Directive, S. F. (2013). Guidance on monitoring of marine litter in European seas. Luxembourg: Publications Office of the European Union, doi, 10, 99475.
18. DJI. (2016). Phantom 4 Pro/Pro+ user manual (Vol. 1.0).
19. DJI. (2018). MAVIC 2 Pro/Zoom User MANUAL (Vol. 1.4): DJI.
20. Duerr, C. (1980). Plastic is forever: our nondegradable treasures. Oceans, November, 1980, 59-60.
21. El Harrouss, O., Moujahid, D., & Tairi, H. (2015). Motion detection based on the combining of the background subtraction and spatial color information. Paper presented at the 2015 Intelligent Systems and Computer Vision (ISCV).
22. Elhabian, S. Y., El-Sayed, K. M., & Ahmed, S. H. (2008). Moving object detection in spatial domain using background removal techniques-state-of-art. Recent patents on computer science, 1(1), 32-54.
23. Fallati, L., Polidori, A., Salvatore, C., Saponari, L., Savini, A., & Galli, P. (2019). Anthropogenic Marine Debris assessment with Unmanned Aerial Vehicle imagery and deep learning: A case study along the beaches of the Republic of Maldives. Sci Total Environ, 693, 133581. doi:10.1016/j.scitotenv.2019.133581
24. Galgani, F., Hanke, G., Werner, S., Oosterbaan, L., Nilsson, P., Fleet, D., . . . Vlachogianni, T. (2013). Monitoring Guidance for Marine Litter in European Seas. JRC Scientific and Policy Reports, Report EUR 26113 EN. In.
25. Gall, S. C., & Thompson, R. C. (2015). The impact of debris on marine life. Marine Pollution Bulletin, 92(1-2), 170-179.
26. Gonçalves, G., Andriolo, U., Gonçalves, L., Sobral, P., & Bessa, F. (2020). Quantifying marine macro litter abundance on a sandy beach using unmanned aerial systems and object-oriented machine learning methods. Remote Sensing, 12(16), 2599.
27. Gonçalves, G., Andriolo, U., Pinto, L., & Bessa, F. (2020). Mapping marine litter using UAS on a beach-dune system: a multidisciplinary approach. Science of The Total Environment, 706, 135742.
28. Gosliner, M. (1984). Legal authorities pertinent to entanglement by marine debris. Paper presented at the Proceedings of the workshop on the fate and impact of marine debris, Honolulu.
29. Goutte, C., & Gaussier, E. (2005). A probabilistic interpretation of precision, recall and F-score, with implication for evaluation. Paper presented at the European conference on information retrieval.
30. Guart, A., Wagner, M., Mezquida, A., Lacorte, S., Oehlmann, J., & Borrell, A. (2013). Migration of plasticisers from Tritan™ and polycarbonate bottles and toxicological evaluation. Food chemistry, 141(1), 373-380.
31. Heyerdahl, T. (1971). Atlantic Ocean pollution and biota observed by the ‘Ra’expeditions. Biological Conservation, 3(3), 164-168.
32. Hoseinnezhad, R., Vo, B.-N., & Vu, T. N. (2011). Visual tracking of multiple targets by multi-Bernoulli filtering of background subtracted image data. Paper presented at the International Conference in Swarm Intelligence.
33. Iñiguez, M. E., Conesa, J. A., & Fullana, A. (2016). Marine debris occurrence and treatment: A review. Renewable and Sustainable Energy Reviews, 64, 394-402. doi:10.1016/j.rser.2016.06.031
34. Intachak, T., & Kaewapichai, W. (2011). Real-time illumination feedback system for adaptive background subtraction working in traffic video monitoring. Paper presented at the 2011 International Symposium on Intelligent Signal Processing and Communications Systems (ISPACS).
35. Jaccard, P. (1901). Distribution de la flore alpine dans le bassin des Dranses et dans quelques régions voisines. Bull Soc Vaudoise Sci Nat, 37, 241-272.
36. Jaccard, P. (1912). The distribution of the flora in the alpine zone. 1. New phytologist, 11(2), 37-50.
37. Joseph, R., & Ali, F. (2017). YOLO9000: Better, Faster, Stronger.
38. Kako, S. i., Morita, S., & Taneda, T. (2020). Estimation of plastic marine debris volumes on beaches using unmanned aerial vehicles and image processing based on deep learning. Marine Pollution Bulletin, 155, 111127.
39. Kamal, A. (2019). Yolo, yolov2 and yolov3: All you want to know. Medium [online].
40. Kataoka, T., Murray, C. C., & Isobe, A. (2018). Quantification of marine macro-debris abundance around Vancouver Island, Canada, based on archived aerial photographs processed by projective transformation. Marine Pollution Bulletin, 132, 44-51.
41. Kylili, K., Kyriakides, I., Artusi, A., & Hadjistassou, C. (2019). Identifying floating plastic marine debris using a deep learning approach. Environmental Science and Pollution Research, 26(17), 17091-17099.
42. Lavers, J. L., Oppel, S., & Bond, A. L. (2016). Factors influencing the detection of beach plastic debris. Marine environmental research, 119, 245-251.
43. Li, G., & Yu, Y. (2016). Deep contrast learning for salient object detection. Paper presented at the Proceedings of the IEEE conference on computer vision and pattern recognition.
44. Liebmann, B., Köppel, S., Königshofer, P., Bucsics, T., Reiberger, T., & Schwabl, P. (2018). Assessment of microplastic concentrations in human stool: Final results of a prospective study. Poster presentation at “Nano and microplastics in technical and freshwater systems–Microplastics, 10.
45. Lippiatt, S., Opfer, S., & Arthur, C. (2013). Marine debris monitoring and assessment: recommendations for monitoring debris trends in the marine environment.
46. Liu, Z., Chen, Z., Li, Z., & Hu, W. (2018). An efficient pedestrian detection method based on YOLOv2. Mathematical Problems in Engineering, 2018.
47. Lo, H. S., Wong, L. C., Kwok, S. H., Lee, Y. K., Po, B. H., Wong, C. Y., . . . Cheung, S. G. (2020). Field test of beach litter assessment by commercial aerial drone. Mar Pollut Bull, 151, 110823. doi:10.1016/j.marpolbul.2019.110823
48. Losada, D. E., & Fernández-Luna, J. M. (2005). Advances in Information Retrieval: 27th European Conference on IR Research, ECIR 2005, Santiago de Compostela, Spain, March 21-23, 2005, Proceedings (Vol. 3408): Springer.
49. Martin, C., Parkes, S., Zhang, Q., Zhang, X., McCabe, M. F., & Duarte, C. M. (2018). Use of unmanned aerial vehicles for efficient beach litter monitoring. Mar Pollut Bull, 131(Pt A), 662-673. doi:10.1016/j.marpolbul.2018.04.045
50. Masura, J., Baker, J., Foster, G., & Arthur, C. (2015). Laboratory Methods for the Analysis of Microplastics in the Marine Environment: Recommendations for quantifying synthetic particles in waters and sediments.
51. Mato, Y., Isobe, T., Takada, H., Kanehiro, H., Ohtake, C., & Kaminuma, T. (2001). Plastic resin pellets as a transport medium for toxic chemicals in the marine environment. Environmental science & technology, 35(2), 318-324.
52. McKee, D., Shuai, B., Berneshawi, A., Wang, M., Modolo, D., Lazebnik, S., & Tighe, J. (2021). Multi-Object Tracking with Hallucinated and Unlabeled Videos. arXiv preprint arXiv:2108.08836.
53. Metz, C. E. (1978). Basic principles of ROC analysis. Paper presented at the Seminars in nuclear medicine.
54. Mikołajczyk, A., & Grochowski, M. (2018). Data augmentation for improving deep learning in image classification problem. Paper presented at the 2018 international interdisciplinary PhD workshop (IIPhDW).
55. Nelms, S., Coombes, C., Foster, L., Galloway, T., Godley, B., Lindeque, P., & Witt, M. (2017). Marine anthropogenic litter on British beaches: a 10-year nationwide assessment using citizen science data. Science of The Total Environment, 579, 1399-1409.
56. Noffke, N. (2010). Geobiology: Microbial mats in sandy deposits from the Archean Era to today: Springer Science & Business Media.
57. Padilla, R., Netto, S. L., & da Silva, E. A. (2020). A survey on performance metrics for object-detection algorithms. Paper presented at the 2020 International Conference on Systems, Signals and Image Processing (IWSSIP).
58. Papakonstantinou, A., Batsaris, M., Spondylidis, S., & Topouzelis, K. (2021). A Citizen Science Unmanned Aerial System Data Acquisition Protocol and Deep Learning Techniques for the Automatic Detection and Mapping of Marine Litter Concentrations in the Coastal Zone. Drones, 5(1), 6.
59. Piccardi, M. (2004). Background subtraction techniques: a review. Paper presented at the 2004 IEEE International Conference on Systems, Man and Cybernetics (IEEE Cat. No. 04CH37583).
60. Pieper, C., Magalhaes Loureiro, C., Law, K. L., Amaral-Zettler, L. A., Quintino, V., Rodrigues, A. M., . . . Martins, A. (2021). Marine litter footprint in the Azores Islands: A climatological perspective. Sci Total Environ, 761, 143310. doi:10.1016/j.scitotenv.2020.143310
61. Pietikäinen, M., Hadid, A., Zhao, G., & Ahonen, T. (2011). Background subtraction. In Computer Vision Using Local Binary Patterns (pp. 127-134): Springer.
62. Powers, D. M. (2007). Evaluation: from precision, recall and F-measure to ROC, informedness, markedness and correlation. Journal of Machine Learning Technologies, 1, 37–63.
63. Ribic, C. A., Dixon, T. R., & Vining, I. (1992). Marine debris survey manual.
64. Risebrough, R. W. (1969, November 23-25, 1969). “The sea: should we now write it off as the future garbage pit?”. Paper presented at the 13th Annual Conference of the U.S. National Commission for UNESCO, San Francisco.
65. Rosevelt, C., Los Huertos, M., Garza, C., & Nevins, H. (2013). Marine debris in central California: Quantifying type and abundance of beach litter in Monterey Bay, CA. Marine Pollution Bulletin, 71(1-2), 299-306. doi:10.1016/j.marpolbul.2013.01.015
66. Sammut, C., & Webb, G. I. (2011). Encyclopedia of machine learning: Springer Science & Business Media.
67. Schulz, M., van Loon, W., Fleet, D. M., Baggelaar, P., & van der Meulen, E. (2017). OSPAR standard method and software for statistical analysis of beach litter data. Mar Pollut Bull, 122(1-2), 166-175. doi:10.1016/j.marpolbul.2017.06.045
68. Shaikh, S. H., Saeed, K., & Chaki, N. (2014). Moving object detection using background subtraction. In Moving object detection using background subtraction (pp. 15-23): Springer.
69. Shevealy, S., Courtney, K., & Parks, J. E. (2012). The Honolulu Strategy: A global framework for prevention and management of marine debris. The Fifth International Marine Debris Conference.
70. Shomura, R. S., & Yoshida, H. O. (1985, 27-29 November 1984). Proceedings of the Workshop on the Fate and Impact of Marine Debris, , Honolulu, Hawaii.
71. Sokolova, M., Japkowicz, N., & Szpakowicz, S. (2006). Beyond accuracy, F-score and ROC: a family of discriminant measures for performance evaluation. Paper presented at the Australasian joint conference on artificial intelligence.
72. Tanimoto, T. (1968). An elementary mathematical theory of classification and prediction, IBM Report (November, 1958), cited in: G. Salton, Automatic Information Organization and Retrieval. In: McGraw-Hill New York.
73. Wagner, M., & Oehlmann, J. (2009). Endocrine disruptors in bottled mineral water: total estrogenic burden and migration from plastic bottles. Environmental Science and Pollution Research, 16(3), 278-286.
74. Wagner, M., & Oehlmann, J. (2011). Endocrine disruptors in bottled mineral water: estrogenic activity in the E-Screen. The Journal of steroid biochemistry and molecular biology, 127(1-2), 128-135.
75. Wehle, D., & Coleman, F. C. (1983). Plastics at sea. Natural History, 92(2), 20.
76. Wenneker, B., & Oosterbaan, L. (2010). Guideline for Monitoring Marine Litter on the Beaches in the OSPAR Maritime Area. Edition 1.0.
77. Wessel, C., Swanson, K., Weatherall, T., & Cebrian, J. (2019). Accumulation and distribution of marine debris on barrier islands across the northern Gulf of Mexico. Mar Pollut Bull, 139, 14-22. doi:10.1016/j.marpolbul.2018.12.023
78. Wilcox, C., Mallos, N. J., Leonard, G. H., Rodriguez, A., & Hardesty, B. D. (2016). Using expert elicitation to estimate the impacts of plastic pollution on marine wildlife. Marine Policy, 65, 107-114.
79. Wright, S. L., & Kelly, F. J. (2017). Plastic and human health: a micro issue? Environmental science & technology, 51(12), 6634-6647.
80. Xue, G., Song, L., Sun, J., & Wu, M. (2011). Hybrid center-symmetric local pattern for dynamic background subtraction. Paper presented at the 2011 IEEE International Conference on Multimedia and Expo.