跳到主要內容

臺灣博碩士論文加值系統

(100.28.227.63) 您好!臺灣時間:2024/06/22 01:34
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果 :::

詳目顯示

: 
twitterline
研究生:張知倫
研究生(外文):Chang, Chih-Lun
論文名稱:以卷積神經網路為基礎作為監測豬隻活動力之工具
論文名稱(外文):Development of a tool for monitoring pig activity based on convolution neural network
指導教授:黃威仁
指導教授(外文):Huang, Wei-Jen
口試委員:楊朝旺蔡循恒
口試日期:2023-08-25
學位類別:碩士
校院名稱:國立嘉義大學
系所名稱:生物機電工程學系
學門:工程學門
學類:機械工程學類
論文種類:學術論文
論文出版年:2023
畢業學年度:111
語文別:中文
論文頁數:69
中文關鍵詞:YOLO精準畜牧業溫溼度指數
外文關鍵詞:You Only Look Once (YOLO)Precision Livestock FarmingpigTemperature-Humidity Index (THI)
相關次數:
  • 被引用被引用:0
  • 點閱點閱:126
  • 評分評分:
  • 下載下載:19
  • 收藏至我的研究室書目清單書目收藏:0
未來全球糧食需求快速增加,畜牧業者須提高動物飼養效率,以符合供不應求的肉品市場。臺灣養豬業屬於集約式畜牧業,需要高度的人力及技術需求,但在高密度飼養及氣候變遷下,動物的健康與福利受到挑戰。精準畜牧業的概念是透過連續性的監控來評估動物的生理及健康狀態,已有許多研究指出,透過感測器監控動物行為變化,有助於發現動物的疾病徵兆或者不適情況。
卷積神經網路(Convolutional Neural Network)屬於人工智慧(Artificial Intelligence)的一種,是一項影像辨識工具,可根據使用者需求,訓練出辨識目標種類的任務,目前最常使用的是YOLO(You Only Look Once)模型。本篇研究中,利用攝影機在豬場端進行影像擷取,並將豬隻行為以二分法區分為趴臥(lying)及非趴臥(not lying),以此進行標記後,使用YOLOv4及YOLOv7模型進行訓練,得到最佳的模型權重及參數,根據本實驗結果,YOLOv4的mAP為0.83,YOLOv7的mAP為0.89。
後續利用YOLOv7模型分析豬場的實際飼養情況以得到豬隻活動情況,並結合溫溼度指數(THI)製成圖表,以此作為監控豬隻活動之工具。以夏季氣候為例,根據分析結果,測試期間的凌晨,THI小於28,豬隻出現非趴臥數量高於趴臥數量;隨著中午的THI達到高峰值31,環境處於嚴重熱緊迫狀態,豬隻多呈現趴臥行為;下午的THI略微降低至28,豬隻的非趴臥行為才又逐漸增加。此活動情形的變化符合文獻對於豬隻行為的研究。
In the face of rapidly increasing global food demand, livestock producers are required to enhance animal farming efficiency to meet the soaring demand for meat products in the market. Taiwan's pig farming industry operates as an intensive livestock system, demanding substantial labor and technological input. However, challenges arise in maintaining animal health and welfare due to high-density rearing practices and the impacts of climate change.
The concept of Precision Livestock Farming involves continuous monitoring to assess the physiological and health status of animals. Many studies have indicated that monitoring behavioral changes through sensors can aid in the detection of disease symptoms or discomfort in animals.
Convolutional Neural Networks (CNN), a subset of Artificial Intelligence, offer powerful tools for image recognition tasks. The YOLO (You-Only-Look-Once) model, particularly YOLOv4 and YOLOv7, is widely used for this purpose. In this study, images of pigs is captured through on-farm cameras. Pig behavior is classified into lying and not-lying categories using binary labeling. Both of models are trained to achieve optimal model weights and parameters. Experimental results reveal a mean Average Precision (mAP) of 0.83 for YOLOv4 and 0.89 for YOLOv7.
Subsequently, the YOLOv7 model is employed to analyze actual rearing conditions, capturing pig activity levels. This data is combined with the Temperature-Humidity Index (THI) to generate graphical representations. Taking the example of summer climate, the analysis results indicate that during the early morning of the testing period when the THI is below 28, there is a higher occurrence of not-lying behavior compared to lying behavior among the pigs. As the THI peaks at 31 around noon, signifying a state of severe heat stress, the pigs predominantly exhibit lying behavior. In the afternoon, with the THI slightly decreasing to 28, not-lying behavior of pigs gradually increases return. This observed pattern of activity variations aligns with existing literature on pig behavior.
摘要 I
Abstract III
致謝 V
目錄 VI
圖目錄 X
表目錄 XII
第一章 前言 1
1-1研究背景 1
1-2 研究目的 3
第二章 文獻探討 4
2-1 臺灣養豬產業之發展與現況 4
2-1-1 口蹄疫疫情過後臺灣養豬型態 4
2-1-2 現代臺灣養豬產業發展 4
2-2 現今畜牧業的發展與趨勢 6
2-2-1 集約式畜牧業與動物健康福利 6
2-2-2 精準畜牧業 7
2-2-3 熱緊迫及溫溼度指數 8
2-3 卷積神經網路的發展 11
2-3-1 人工智慧Artificial Intelligence 11
2-3-2 機器學習 Machine Learning 12
2-3-3 深度學習 Deep Learning 12
2-3-4 卷積神經網路 Convolutional Neural Network 12
2-4 畜牧業中的影像辨識 14
第三章 基礎理論 17
3-1 卷積神經網路 17
3-1-1卷積層(Convolution Layer) 17
3-1-2 池化層(Pooling Layer) 18
3-1-3 周圍補充(Padding) 19
3-1-4 激勵函數(Activation Function) 20
3-1-5 交並比(Intersection over Union; IoU) 21
3-1-6 非極大值抑制(Non-Max Suppression; NMS) 22
3-1-7 準確率(mean Average Precision ;mAP ) 23
3-2 YOLO神經網路 25
3-2-1 YOLOv4 27
3-2-2 YOLOv7 29
第四章 材料與方法 31
4-1 實驗架構 31
4-2 動物與實驗場所 33
4-3 收集訓練用資料 33
4-4 訓練模型 36
4-4-1 標記影像特徵 36
4-4-2 訓練神經網路模型 38
4-4-3 選擇最佳權重 41
4-5批次產生影像辨識結果 41
4-6 溫溼度指數 43
4-7 豬隻活動力監測 43
第五章 結果與討論 45
5-1 YOLOv4模型訓練結果 45
5-2 YOLOv7模型訓練結果 48
5-3 過度擬合(Overfitting) 52
5-4 模型效能比較 53
5-4-1 Faster R-CNN模型與YOLO模型之比較 53
5-4-2 YOLO模型之比較 55
5-5 豬隻活動力監測 57
5-6 豬隻行為辨識模型的改良 60
第六章 結論與建議 62
6-1結論 62
6-2建議 63
參考文獻 64
王斌永、賴佑宜、蔡銘洋。2021。豬場智慧化防疫行動管理系統開發。畜產研究 54(2): 99-105。
中華民國審計部全球資訊網。2023。網址:https://www.audit.gov.tw/
行政院農業委員會。2023。網址:https://www.moa.gov.tw/index.php
黃裕益、謝廣文、鄭俊哲。2000。可程式邏輯控制器於水濂式畜舍環控管理之研究。畜產研究 33(4): 384-396。
黃偉豪。2022。應用深度學習技術於蘆筍生長階段及病害辨識之應用。碩士論文。臺北:臺灣大學生物機電工程學系研究所。
張哲維。2021。以基於多重偵測與目標檢測演算法YOLO實現禽類健康偵測和辨識。碩士論文。嘉義:國立嘉義大學資訊工程學系研究所。
郭金江、雷鵬魁、沈葆南。2008。噴霧系統應用於養豬場與堆肥場除臭之研究。農業機械學刊 17(3): 25-38。
劉志偉。2009。國際農糧體制與臺灣的糧食依賴:戰後臺灣養豬業的歷史考察。《臺灣史研究》 16(2): 105-160。
蔡侑容。2022。利用卷積神經網路監測分娩欄位內母豬及仔豬泌乳相關行為。碩士論文。臺北:臺灣大學生物機電工程學系研究所。
廖冠瑋。2022。利用深度學習與水下影像研發白蝦成長及行為模式的物聯網監測系統。碩士論文。臺北:臺灣大學生物機電工程學系研究所。
Achour, B., M. Belkadi, I. Filali, M. Laghrouche, and M. Lahdir. 2020. Image analysis for individual identification and feeding behaviour monitoring of dairy cows based on Convolutional Neural Networks (CNN). Biosystems Engineering 198, 31-49.
Alameer, A., I. Kyriazakis, and J. Bacardit. 2020. Automated recognition of postures and drinking behaviour for the detection of compromised health in pigs. Scientific reports 10(1): 13665.
Andrew, W., C. Greatwood, and T. Burghardt. 2019. Aerial animal biometrics: Individual friesian cattle recovery and visual identification via an autonomous uav with onboard deep inference. In 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 237-243.
Arago, N. M., C. I. Alvarez, A. G. Mabale, C. G. Legista, N. E. Repiso, R. R. A. Robles, T. M. Amado, R. L. Jorda, A. C. Thio-ac, J. S. Velasco and L. K. S. Tolentino. 2020. Automated estrus detection for dairy cattle through neural networks and bounding box corner analysis. International Journal of Advanced Computer Science and Applications 11(9).
Banhazi, T. 2013. Environmental and management effects associated with improved production efficiency in a respiratory disease free pig herd in Australia. Livestock housing 49-56.
Bergeron, R., A. J. Badnell-Waters, S. Lambton and G. Mason. 2006. Stereotypic oral behaviour in captive ungulates: foraging, diet and gastrointestinal function. In Stereotypic animal behaviour: Fundamentals and applications to welfare, 19-57. Wallingford UK: CABI.
Bochkovskiy, A., C. Y. Wang and H. Y. Liao. 2020. Yolov4: Optimal speed and accuracy of object detection. arXiv preprint arXiv: 2004.10934.
Chantziaras, I., D. De Meyer, L. Vrielinck, T. Van Limbergen, C. Pineiro, J. Dewulf, I. Kyriazakis and D. Maes. 2020. Environment-, health-, performance-and welfare-related parameters in pig barns with natural and mechanical ventilation. Preventive veterinary medicine 183, 105150.
Cross, A. J., T. M. Brown-Brandl, B. N. Keel, J. P. Cassady and G. A. Rohrer, G. A. 2020. Feeding behavior of grow-finish swine and the impacts of heat stress. Translational Animal Science 4(2): 986-992.
Ewbank, R. 1976. Social hierarchy in suckling and fattening pigs: a review. Livestock Production Science 3(4): 363-372.
Fitzgerald, R. F., K. J. Stalder, J. O. Matthews, C. M. Schultz Kaster and A. K. Johnson. 2009. Factors associated with fatigued, injured, and dead pig frequency during transport and lairage at a commercial abattoir. Journal of animal science 87(3): 1156-1166.
Hälli, O., O. A. T. Peltoniemi, A. Tast, J. V. Virolainen, C. Munsterhjelm, A. Valros and M. Heinonen. 2008. Photoperiod and luteinizing hormone secretion in domestic and wild pigs. Animal reproduction science 103(1-2): 99-106.
Hinton, G. E., S. Osindero and Y. W. Teh. 2006. A fast learning algorithm for deep belief nets. Neural computation 18(7): 1527-1554.
Jiang, M., Y. Rao, J. Zhang and Y. Shen. 2020. Automatic behavior recognition of group-housed goats using deep learning. Computers and Electronics in Agriculture 177: 105706.
Lallo, C. H., J. Cohen, D. Rankine, M. Taylor, J. Cambell and T. Stephenson. 2018. Characterizing heat stress on livestock using the temperature humidity index (THI)—prospects for a warmer Caribbean. Regional Environmental Change 18: 2329-2340.
LeCun, Y., B, Boser, J. S. Denker, D. Henderson, R. E. Howard, W. Hubbard and L. D. Jackel. (1989). Backpropagation applied to handwritten zip code recognition. Neural Computation 1(4): 541-551
Madsen, T. N. and A. R. Kristensen. 2005. A model for monitoring the condition of young pigs by their drinking behaviour. Computers and electronics in agriculture 48(2): 138-154.
Maselyne, J., M. Pastell, P. T. Thomsen, V. M. Thorup, L. Hänninen, J. Vangeyte, A. V. Nuffel and L. Munksgaard. 2017. Daily lying time, motion index and step frequency in dairy cows change throughout lactation. Research in veterinary science 110: 1-3.
Matthews, S. G., A. L. Miller, J. Clapp, T. Plötz and I. Kyriazakis. 2016. Early detection of health and welfare compromises through automated detection of behavioural changes in pigs. The Veterinary Journal 217: 43-51.
McLeod, A. 2011. World livestock 2011-livestock in food security. Food and agriculture organization of the united nations (FAO).
Moorby, J. M. and M. D. Fraser. 2021. New feeds and new feeding systems in intensive and semi-intensive forage-fed ruminant livestock systems. Animal 15: 100297.
Nyachoti, C. M., R. T. Zijlstra, D. F. M. De Lange and J. F. Patience. 2004. Voluntary feed intake in growing-finishing pigs: A review of the main determining factors and potential approaches for accurate predictions. Canadian journal of animal science 84(4): 549-566.
Renaudeau, D., G. Francès, S. Dubois, H. Gilbert and J. Noblet. 2013. Effect of thermal heat stress on energy utilization in two lines of pigs divergently selected for residual feed intake. Journal of animal science 91(3): 1162-1175.
Riekert, M., A. Klein, F. Adrion, C. Hoffmann and E. Gallmann. 2020. Automatically detecting pig position and posture by 2D camera imaging and deep learning. Computers and Electronics in Agriculture 174: 105391.
Rodriguez Alvarez, J., M. Arroqui, P. Mangudo, J. Toloza, D. Jatip, J. M. Rodriguez, S. Teyseyre, Z. Alejandro, M. Claudio and C. Mateos. 2019. Estimating body condition score in dairy cows from depth images using convolutional neural networks, transfer learning and model ensembling techniques. Agronomy 9(2): 90.
Rumelhart, D. E., G. E. Hinton and R. J. Williams. 1986. Learning representations by back-propagating errors. nature 323(6088): 533-536.
Terven, J. and D. M. Cordova-Esparza. 2023. A Comprehensive Review of YOLO: From YOLOv1 and Beyond. arXiv preprint arXiv: 2304.00501.
Vranken, E. and D. Berckmans. 2017. Precision livestock farming for pigs. Animal Frontiers 7(1): 32-37.
Wang, C. Y., A. Bochkovskiy and H. Y. Liao. 2023. YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition 7464-7475.
Weary, D. M., J. M. Huzzey and M. A. Von Keyserlingk. 2009. BOARD-INVITED REVIEW: Using behavior to predict and identify ill health in animals1. Journal of Animal Science 87(2): 770-777.
Yang, A., H. Huang, C. Zheng, X. Zhu, X. Yang, P. Chen and Y. Xue. 2018. High-accuracy image segmentation for lactating sows using a fully convolutional network. Biosystems Engineering 176: 36-47.
Yitbarek, M. B. 2019. Livestock and livestock product trends by 2050. Italian Journal of Addiction Reaserch 4: 30.
連結至畢業學校之論文網頁點我開啟連結
註: 此連結為研究生畢業學校所提供,不一定有電子全文可供下載,若連結有誤,請點選上方之〝勘誤回報〞功能,我們會盡快修正,謝謝!
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top