- 學位論文
皮革瑕疵之AI深度學習技術導入研究
Study of Techniques for AI Deep Learning Leather Defects
若您是本文的作者，可授權文章由華藝線上圖書館中協助推廣。
摘要
人造皮革在製造及加工過程時會有表面瑕疵的產生，一般由作業員肉眼檢測，由於人造皮革的種類、顏色和花紋非常多樣化，加上瑕疵類型也非常多種，以人工方式進行瑕疵檢測，長時間下很難維持一定的準確性，使得不僅檢測人力成本較高，檢測效率與品質也不穩定。因此，本研究的主要目標是運用人工智慧深度學習技術進行人造皮革瑕疵檢測，藉此確認技術的可行性。 本研究透過產學合作計畫，由公司提供的皮革瑕疵樣品，經過皮革瑕疵取像、AI深度學習模型(YOLOv3)的建立與訓練的可行性分析，最後對深度學習模型的準確率、精準率和召回率進行的評估分析。實驗結果中，總共拍攝7,167張圖像用來進行深度學習模型的訓練與測試，最終透過四個模型的組合，使得準確率可以達到89.7%，證明深度學習技術在人造皮革的檢測上具有一定的可行性。
並列摘要
The manufacturing of artificial leather may causes various defects on the surface. Visual inspection of operations is a common way. The types, colors, or textures of artificial leathers are diverse. There are also many kinds of defects. The long-time inspection of operators is difficult to keep a stable accuracy. The cost of inspection process is high. The inspection efficiency and quality is not good. Therefore, the purpose of this study is the integration of deep learning technique to evaluate the feasibility of using the technique on artificial leather defect inspection. This study is mainly based on an academic-industrial cooperative project. The company provides the leathers with defects. The defect images of these leathers are captured and used in training and testing of the deep learning model (YOLOv3). The feasibility evaluation is according to the accuracy, precision and recall of the deep learning model. In the experimental results, 7,167 images are captured for training and testing of the deep learning model. The final accuracy of the combination of four models can achieve 89.7%. It shows that the deep learning technique is feasible for the artificial leather defect inspection.
參考文獻
參考文獻
[1] M. Dill, “Leather,” Wikipedia[Online]. Available: https://en.wikipedia.org/wiki/Leather, 14 July 2020.
[2] M. K. Kasi, J B. Rao and V. K. Sahu, “Identification of Leather Defects using An Autoadaptive Edge Detection Image Processing Algorithm,” 2014 International Conference on High Performance Computing and Applications (ICHPCA), 19 February 2015.
[3] T.-H. H. Lee, “Edge Detection Analysis,” [Online]. Available: http://disp.ee.ntu.edu.tw/henry/edge_detection.pdf, 27 May 2020.
[4] T. Qiu, Y. Yan, G. Lu, “An Autoadaptive Edge-Detection Algorithm for Flame and Fire Image Processing,” IEEE Transactions on Instrumentation and Measurement, VOL. 61, NO. 5, MAY 2012.