完整後設資料紀錄
DC 欄位語言
dc.contributor.author陳麗妃zh_TW
dc.contributor.author黃韻雅zh_TW
dc.contributor.author張麗雪zh_TW
dc.contributor.author蘇朝墩zh_TW
dc.contributor.authorLi-Fei Chenen_US
dc.contributor.authorYun-Ya Huangen_US
dc.contributor.authorLi-Hsueh Changen_US
dc.contributor.authorChao-Ton Suen_US
dc.date.accessioned2022-12-19T08:08:10Z-
dc.date.available2022-12-19T08:08:10Z-
dc.date.issued2022-04-01en_US
dc.identifier.issn1023-9863en_US
dc.identifier.urihttp://dx.doi.org/10.29416/JMS.202204_29(2).0004en_US
dc.identifier.urihttp://hdl.handle.net/11536/159690-
dc.description.abstract隨著工業技術的發展,工廠通常導入自動光學檢查系統,以識別生產線中的產品缺陷。但是,訓練圖像分類系統仍然存在一些困難,例如數據不平衡和費時的標記。為了解決這些問題,本研究透過使用卷積神經網絡、遷移學習和資料擴增方法,提出了一有效的缺陷分類建模程序。使用台灣某LED顯示器製造商提供的LED板材瑕疵影像資料來構建模型,並分析模型的準確性和訓練時間。在進行比較分析之後,本研究建議使用線上資料擴增手法來增加資料集的變化性,然後再微調預訓練模型以進行瑕疵分類,該模型分類準確度可高達98%。此結果說明,使用所提出的程序,工廠可以快速建立缺陷分類系統以監視生產流程。zh_TW
dc.description.abstractWith the development of industrial technology, it is common for factories to introduce automatic optical inspection systems to identify defects in production lines. However, there are still some difficulties in training image classification systems, such as unbalanced data and time-consuming labeling. In order to solve these problems, this study proposes an efficient defect classification modeling procedure by using convolutional neural networks, transfer learning, and data augmentation methods. LED lead frame defect images provided by a Taiwanese LED display manufacturer were used to construct the models and analyze model accuracy and training time. Finally, following comparative analysis, this study suggests applying an online data augmentation method to increase the variability of the dataset and then fine-tuning the pre-trained model to learn and classify the defect features; the classification accuracy of the proposed model can be as high as 98%. This result shows that, using the proposed procedure, a factory can quickly establish a defect classification system to monitor the production process. 隨著工業技術的發展,工廠通常導入自動光學檢查系統,以識別生產線中的產品缺陷。但是,訓練圖像分類系統仍然存在一些困難,例如數據不平衡和費時的標記。為了解決這些問題,本研究透過使用卷積神經網絡、遷移學習和資料擴增方法,提出了一有效的缺陷分類建模程序。使用台灣某LED顯示器製造商提供的LED板材瑕疵影像資料來構建模型,並分析模型的準確性和訓練時間。在進行比較分析之後,本研究建議使用線上資料擴增手法來增加資料集的變化性,然後再微調預訓練模型以進行瑕疵分類,該模型分類準確度可高達98%。此結果說明,使用所提出的程序,工廠可以快速建立缺陷分類系統以監視生產流程。en_US
dc.language.isoen_USen_US
dc.publisher國立陽明交通大學經營管理研究所zh_TW
dc.publisherInstitute of Business and Magement, National Yang Ming Chiao Tung Universityen_US
dc.subject卷積神經網路zh_TW
dc.subject遷移學習zh_TW
dc.subject資料擴增zh_TW
dc.subject瑕疵分類zh_TW
dc.subjectCNNen_US
dc.subjectTransfer Learningen_US
dc.subjectData Augmentationen_US
dc.subjectDefect Classificationen_US
dc.title運用卷積神經網路遷移學習與資料擴增方法於瑕疵之分類zh_TW
dc.titleDefect Classification Using CNN Transfer Learning and Data Augmentationen_US
dc.typeCampus Publicationsen_US
dc.identifier.doi10.29416/JMS.202204_29(2).0004en_US
dc.identifier.journal管理與系統zh_TW
dc.identifier.journalJournal of Management and Systemsen_US
dc.citation.volume29en_US
dc.citation.issue2en_US
dc.citation.spage223en_US
dc.citation.epage239en_US
顯示於類別:管理與系統