論文使用權限 Thesis access permission:自定論文開放時間 user define
開放時間 Available:
校內 Campus: 已公開 available
校外 Off-campus: 已公開 available
論文名稱 Title |
基於深度卷積自編碼的圖像檢索系統 Image Retrieval Based On Deep Convolutional Autoencoders |
||
系所名稱 Department |
|||
畢業學年期 Year, semester |
語文別 Language |
||
學位類別 Degree |
頁數 Number of pages |
41 |
|
研究生 Author |
|||
指導教授 Advisor |
|||
召集委員 Convenor |
|||
口試委員 Advisory Committee |
|||
口試日期 Date of Exam |
2019-07-22 |
繳交日期 Date of Submission |
2019-08-27 |
關鍵字 Keywords |
距離演算法、數據降維、自編碼、卷積神經網路、深度學習 Distance Algorithm, Data Dimensionality Reduction, Deep Learning, Convolutional Neural Network, Autoencoder |
||
統計 Statistics |
本論文已被瀏覽 5974 次,被下載 72 次 The thesis/dissertation has been browsed 5974 times, has been downloaded 72 times. |
中文摘要 |
電腦視覺圖像辨識技術,受惠於近幾年來深度學習各類演算法的發展演進,搭配GPU運算能力的支援,在天時地利的情況下有了不同於傳統只能辨識簡單圖形的能力,通過深度學習的強大學習能力,圖像辨識的能力及精確度,已接近甚至超越人類所能,足以協助人類處理圖像辨識的工作。 本研究目的是圖像檢索系統,基於深度學習的方式,使用卷積自編碼神經網路並引用Stanford Dogs Dataset 和UECFOOD256 Dataset等不同類型資料集,先進行自編碼模型的訓練,再利用自編碼模型中的編碼器進行圖像特徵提取,再將特徵數據降維後,通過距離演算法的計算找出特徵近似的圖像,提出一套最可行的圖像檢索系統。 |
Abstract |
Computer vision image recognition technology has benefited from the development of various algorithms in deep learning in recent years. With the support of GPU computing power, through the powerful learning ability of deep learning, the ability and accuracy of image recognition is close to or beyond human ability, enough to assist humans in the work of image recognition. The purpose of this study is an image retrieval system, based on deep learning, using a convolutional autoencoder neural network and citing different types of data sets such as Stanford Dogs Dataset and UECFOOD256 Dataset. First training the autoencoder model, and then using the encoder extract the image features. After reducing the dimensionality of features data, the image of the feature approximation is found by the distance computation. |
目次 Table of Contents |
論文審定書 i 摘要 ii Abstract iii 誌謝 iv 目錄 v 圖次 vii 表次 viii 1 緒論 1 1.1 研究背景 1 1.2 研究動機 1 1.3 研究目的 2 2 文獻探討及相關研究 3 2.1 深度學習 3 2.2 卷積神經網路(Convolutional Neural Network) 5 2.2.1 卷積層(Convolutional Layer) 6 2.2.2池化層(Pooling Layer) 7 2.2.3完全連接層(Full connect Layer) 8 2.3 自編碼(Autoencoder) 9 3 圖像檢索系統研究與實作方法 10 3.1 研究方法 10 3.1.1 PCA(Principal Component Analysis) 11 3.1.2 t-SNE (t-distributed Stochastic Neighbor Embedding) 12 3.1.3距離演算法 13 3.1.3.1 歐基里德距離(Euclidean Distance) 13 3.1.3.2 城市街區距離(Cityblock Distance) 14 3.1.3.3夾角餘弦(Cosine Distance) 15 3.1.3.4 相關距離(Correlation Distance) 16 3.1.3.5 BrayCurtis 距離(Braycurtis Distance) 16 3.1.3.6坎培拉距離(Canberra Distance) 17 3.1.3.7切比雪夫距離(Chebyshev Distance) 18 3.2 使用預訓練模型實作 18 3.3 深度卷積自編碼實作 20 4 實驗結果及討論 22 4.1 實驗結果 22 4.2 討論 29 5 參考文獻 31 |
參考文獻 References |
Abdi, H., & J. Williams, L. (2010). Principal component analysis. (Computational Statistics 2), 433–459. Chollet, F. (2017). Deep Learning with Python. Chollet, F. (n.d.). Building Autoencoders in Keras. Retrieved from https://blog.keras.io/building-autoencoders-in-keras.html CS231n Convolutional Neural Networks for Visual Recognition. (n.d.). Retrieved from http://cs231n.github.io/convolutional-networks/ Distance computations. (n.d.). Retrieved from https://docs.scipy.org/doc/scipy/reference/spatial.distance.html Hinton, G. E., & Salakhutdinov, R. R. (2006). Reducing the Dimensionality of Data with Neural Networks. (Science,313 5786), 504–507. How do Convolutional Neural Networks work? (n.d.). Retrieved from https://brohrer.github.io/how_convolutional_neural_networks_work.html J. Deng, W. Dong, R. Socher, L. Li, Kai Li, & Li Fei-Fei. (2009). ImageNet: A large-scale hierarchical image database. 2009 IEEE Conference on Computer Vision and Pattern Recognition, 248–255. https://doi.org/10.1109/CVPR.2009.5206848 Kawano, Y., & Yanai, K. (2015). Automatic Expansion of a Food Image Dataset Leveraging Existing Categories with Domain Adaptation. In L. Agapito, M. M. Bronstein, & C. Rother (Eds.), Computer Vision - ECCV 2014 Workshops (pp. 3–17). Springer International Publishing. Khosla, A., Jayadevaprakash, N., Yao, B., & Li, F.-F. (2011). Novel Dataset for Fine-Grained Image Categorization: Stanford Dogs. (IEEE Conference on Computer Vision and Pattern Recognition (CVPR)). LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521, 436. LeCun, Y., Cortes, C., & J.C. Burges, C. (2005). THE MNIST DATABASE of handwritten digits. Retrieved from http://yann. lecun. com/exdb/mnist/ Lisa Torrey, & Jude Shavlik. (2010). Transfer Learning. In Emilio Soria Olivas, José David Martín Guerrero, Marcelino Martinez-Sober, Jose Rafael Magdalena-Benedito, & Antonio José Serrano López (Eds.), Handbook of Research on Machine Learning Applications and Trends: Algorithms, Methods, and Techniques (pp. 242–264). https://doi.org/10.4018/978-1-60566-766-9.ch011 Mainfold learning. (n.d.). Retrieved from https://scikit-learn.org/stable/modules/manifold.html#manifold Suárez-Paniagua, V., & Segura-Bedmar, I. (2018). Evaluation of pooling operations in convolutional architectures for drug-drug interaction extraction. BMC Bioinformatics, 19(8), 209. https://doi.org/10.1186/s12859-018-2195-1 van der Maaten, L., & Hinton, G. (n.d.). Visualizing Data using t-SNE. (Journal of Machine Learning Research 9 (2008)), 2579–2605. Zacharski, R. (n.d.). A Programmer’s Guide to Data Mining. Retrieved from https://www.freetechbooks.com/a-programmers-guide-to-data-mining-the-ancient-art-of-the-numerati-t925.html |
電子全文 Fulltext |
本電子全文僅授權使用者為學術研究之目的,進行個人非營利性質之檢索、閱讀、列印。請遵守中華民國著作權法之相關規定,切勿任意重製、散佈、改作、轉貼、播送,以免觸法。 論文使用權限 Thesis access permission:自定論文開放時間 user define 開放時間 Available: 校內 Campus: 已公開 available 校外 Off-campus: 已公開 available |
紙本論文 Printed copies |
紙本論文的公開資訊在102學年度以後相對較為完整。如果需要查詢101學年度以前的紙本論文公開資訊,請聯繫圖資處紙本論文服務櫃台。如有不便之處敬請見諒。 開放時間 available 已公開 available |
QR Code |