臺灣博碩士論文加值系統

深度學習在影像分類、辨識方面比傳統技術有著優秀的表現，在Hinton等人利用卷積神經網路來提高影像辨識準確率後，諸多學者及公司便開始著手於深度學習模組的開發，如AlexNet、VGGNet、Google Inception Net等等，本研究主要使用Google Inception Net 所提供Inception模組來訓練自己的圖庫，我們會調整Inception模組裡包括學習率、訓練疊代次數及Inception的模組版本為實驗參數來提高模組的準確率。在實驗中會使用ImageNet中的貓、狗、青蛙、鹿、鳥等五種動物作為測試圖庫，我們將使用影像前處理演算法來進行固定閥值的二值化處理讓圖像感性區域更明顯，再以原始圖庫和二值化過後的圖庫作為模組的訓練集，而模組版本將會以Google Inception的Inception V1、Inception V2與Inception V3等三個版本來進行訓練，其模型訓練參數包括學習率為0.01、0.005及0.001，訓練疊代次數為4000、3000、2000及1000等，最後會以準確率最高的結果透過本文所選用的六種匹配模組來成功圈出測試目標及貼上標籤。

Deep learning has excellent performance in image classification and recognition compared with traditional technology. After Hinton et al. used convolutional neural networks to improve image recognition accuracy, many scholars and companies began to develop deep learning modules, such as AlexNet, VGGNet, Google Inception Net, etc. This study mainly uses the Inception module provided by Google Inception Net to train its own images. We will adjust the Inception module to include the learning rate, the number of training iterations and the module version of Inception. Experimental parameters to improve the accuracy of the module. In the experiment, we will use five kinds of animals such as cats, dogs, frogs, deer and birds in ImageNet as the test image. We will use the image pre-processing algorithm to carry out the binarization of the fixed threshold to make the image inductive area more obvious. The original library and the binarized gallery are used as the training set of the module, and the module version will be trained by three versions of Google Inception Inception V1, Inception V2 and Inception V3. The model training parameters include The learning rate is 0.01, 0.005 and 0.001, and the number of training iterations is 4000, 3000, 2000 and 1000, etc.Finally, the highest accuracy results will be used to successfully circle the test target and label it through the six matching modules selected in this article.

摘要
Abstract
誌謝
目錄
表目錄
圖目錄
一、 緒論
1.1 研究動機
1.2 論文架構
二、 文獻回顧
2.1 TensorFlow
2.2 ImageNet
2.3 Google Inception Net
2.4 OpenCV
2.5 機器學習
2.6 深度學習
2.7 類神經網路
2.8 卷積神經網路
三、 研究方法
3.1 Google Inception Net模型架構
3.2 資料前處理
3.3 Google Inception Net參數設定
3.4 匹配模組
四、 研究結果與探討
4.1 研究結果
五、結論與未來展望
參考文獻

[1]A. M. Turing, "Computing Machinery and Intelligence. Mind 49: 433-460.",950
[2]Searle, John R. "Minds, brains, and programs." Behavioral and brain sciences 3.3 (1980): 417-424
[3]Holmes, R. J., et al. "Efficient, deep-blue organic electrophosphorescence by guest charge trapping." Applied Physics Letters 83.18 (2003): 3818-3820.
[4]Y. LeCun, Y. Bengio, and G. Hinton, ‘‘ Deep learning,’’ Nature, 521(7553), 436-444, 2015.
[5]Silver, D. et al.,“Mastering the Game of Go with Deep Neural Networks and Tree Search”, Nature, vol. 529, pp. 484-489, 2016.
[6]Krizhevsky, Alex, Ilya Sutskever, and Geoffrey E. Hinton. "Imagenet classification with deep convolutional neural networks." Advances in neural information processing systems. 2012.
[7]Simonyan, Karen, and Andrew Zisserman. "Very deep convolutional networks for large-scale image recognition." arXiv preprint arXiv:1409.1556 (2014).
[8]Szegedy, Christian, et al. "Going deeper with convolutions." Proceedings of the IEEE conference on computer vision and pattern recognition. 2015.
[9]Deng, Jia, et al. "Imagenet: A large-scale hierarchical image database." Computer Vision and Pattern Recognition, 2009. CVPR 2009. IEEE Conference on. Ieee, 2009.
[10]Ioffe, Sergey, and Christian Szegedy. "Batch normalization: Accelerating deep network training by reducing internal covariate shift." arXiv preprint arXiv:1502.03167 (2015).
[11]Szegedy, Christian, et al. "Rethinking the inception architecture for computer vision." Proceedings of the IEEE conference on computer vision and pattern recognition. 2016.
[12]Abadi, Martín, et al. "Tensorflow: a system for large-scale machine learning." OSDI. Vol. 16. 2016.
[13]Russakovsky, Olga, et al. "Imagenet large scale visual recognition challenge." International Journal of Computer Vision 115.3 (2015): 211-252.
[14]Yegnanarayana, B. Artificial neural networks. PHI Learning Pvt. Ltd., 2009.
[15]Russell, Stuart J., and Peter Norvig. Artificial intelligence: a modern approach. Malaysia; Pearson Education Limited,, 2016.
[16]Hastie, Trevor, Robert Tibshirani, and Jerome Friedman. "Unsupervised learning." The elements of statistical learning. Springer, New York, NY, 2009. 485-585.
[17]Sutton, Richard S., Andrew G. Barto, and Francis Bach. Reinforcement learning: An introduction. MIT press, 1998.
[18]A. Criminisi, ‘‘Machine learning for medical images analysis,’’ 2016.
[19]A. Oishi and G. Yagawa, ‘‘Computational mechanics enhanced by deep learning,’’ Computer Methods in Applied Mechanics and Engineering, 2017.
[20]M. Sun, Z. Song, X. Jiang, J. Pan, and Y. Pang, ‘‘Learning Pooling for Convolutional Neural Network,’’ Neurocomputing, 224, 96-104, 2017.
[21]C. Affonso, A. L. D. Rossi, F. H. A. Vieira, and A. C. P. de Leon Ferreira de Carvalho, ‘‘Deep learning for biological image classification,’’ Expert Systems with Applications, 85, 114-122, 2017.
[22]D. Norris, ‘‘Shortlist: A connectionist model of continuous speech recognition,’’ Cognition, 52(3), 189-234, 1994.
[23]G. G. Chowdhury, ‘‘Natural language processing,’’ Annual review of information science and technology, 37(1), 51-89, 2003.
[24]Bottou, Léon. "Large-scale machine learning with stochastic gradient descent." Proceedings of COMPSTAT'2010. Physica-Verlag HD, 2010. 177-186.
[25]黃文堅、唐源，2017，實戰 TensorFlow Google 深度學習，蔡勝文，初版，碁峰資訊股份有限公司，台北市
[26]斎藤康毅，2017，Deep Learning :用python 進行深度學習的基礎理論實作，吳嘉芳，初版，碁峰資訊股份有限公司，台北市
[27]林大貴，2017，TensorFlow + Keras 深度學習人工智慧實務應用，初版，博碩文化股份有限公司，新北市
[28]Nikhil Buduma ，2018，Deep Learning 深度學習基礎 :設計下一代人工智慧演算法，藍子軒，初版，碁峰資訊股份有限公司，台北市
[29]高揚，2018，白話深度學習與TensorFlow，初版，碁峰資訊股份有限公司，台北市
[30]何之源，2018，學會TensorFlow之後呢? 最尖端的21個夢幻範例實作，初版，家魁數位，臺北市
[31]OpenCV二值化官網 https://ppt.cc/fkTqux
[32]OpenCV匹配模組官網https://ppt.cc/f7OGmx