隨著科技發展，人工智慧已經成為我們生活中不可或缺的一部分，比如手機上各式各樣的應用服務結合了人工智慧，讓使用者更加便利的使用，它不僅改善了生活品質，也提高了工作效率。傳統的機器學習通常會使用高階的計算機來進行應用，但在使用者操作時可能不需要用到這麼高階的計算機，所以就產生了如果可以把機器學習運用在更低階的計算機、準確率稍微降低一點也沒有問題，這種情況下的想法。
本論文使用機器學習技術，以自行設計的小型卷機神經模型(稱為MyCNN)、MobilenetV1和Tesseract-OCR三種模型，來實現車牌上的數字與英文字母的辨識。除了在電腦（PC）上辨識，另外，模型進一步通過量化(Quantization) 技術來降低模型精度，然後在樹莓派（Raspberry Pi）硬體平台實現嵌入式環境中的辨識。最終，本研究探討這三種模型在正確率和辨識速度方面的表現，透過比較在電腦和樹莓派這兩種硬體環境下的實驗結果，希望提供方便且準確的預測。

With the advancement of technology, artificial intelligence has become an indispensable part of our lives. For example, various application services on smartphones integrate artificial intelligence, providing users with greater convenience. It not only improves the quality of life but also enhances work efficiency. Traditional machine learning typically involves the use of high-end computers for applications. However, during user interactions, there may be no need for such advanced computing power. This has led to the idea that if machine learning can be applied to lower-level computers, with a slight decrease in accuracy, it would still be acceptable in certain scenarios.
This thesis employs machine learning techniques, utilizing three models: a custom-designed small convolutional neural network (referred to as MyCNN), MobilenetV1, and Tesseract-OCR. The objective is to achieve the recognition of numbers and English letters on license plates. In addition to recognition on a computer (PC), the models undergo further quantization techniques to reduce precision, facilitating recognition in an embedded environment on the Raspberry Pi hardware platform. Ultimately, this study explores the performance of these three models in terms of accuracy and recognition speed. By comparing the experimental results on both computer and Raspberry Pi hardware environments, the goal is to provide a convenient and accurate prediction for license plate recognition.

摘要 I
Abstract II
目錄 IV
圖目錄 V
表目錄 VI
第一章、緒論 1
1.1 前言 1
1.2 研究動機與目的 1
1.3 論文組織架構 2
第二章、研究背景 3
2.1 人工智慧、機器學習、深度學習 3
2.2 TinyML 6
2.3 TensorFlow和TensorFlow Lite 6
2.3.1 TensorFlow簡介 7
2.3.2 TensorFlow Lite簡介 8
2.3.3 TensorFlow與TensorFlow Lite的應用 8
2.4 MobileNet介紹 9
2.5 Tesseract OCR 11
2.6 車牌辨識相關技術 13
2.7 AOLP 13
第三章、車牌辨識系統架構與處理流程 15
3.1 系統架構 15
3.2 車牌辨識實作流程 16
3.3 模型建立 18
第四章、實驗與結果分析 21
4.1 實驗環境 21
4.2 測試方法 21
4.3 實驗結果展示與分析 23
第五章、結論 79
參考文獻 80

[1] Helm, J.M., Swiergosz, A.M., Haeberle, H.S., Karnuta, J.M., Schaffer, J.L., Krebs, V.E., Spitzer, A.I., & Ramkumar, P.N. (2020). Machine Learning and Artificial Intelligence: Definitions, Applications, and Future Directions. Current Reviews in Musculoskeletal Medicine, 13, 69-76.
[2] Sarker, I.H. (2021). Deep Learning: A Comprehensive Overview on Techniques, Taxonomy, Applications and Research Directions. Sn Computer Science, 2.
[3] Nwankpa, C., Ijomah, W.L., Gachagan, A., & Marshall, S. (2018). Activation Functions: Comparison of trends in Practice and Research for Deep Learning. ArXiv, abs/1811.03378.
[4] Barron, J.T. (2017). A General and Adaptive Robust Loss Function. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 4326-4334.
[5] Albawi, S., Mohammed, T.A., & Al-Zawi, S. (2017). Understanding of a convolutional neural network. 2017 International Conference on Engineering and Technology (ICET), 1-6.
[6] Sherstinsky, A. (2018). Fundamentals of Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM) Network. ArXiv, abs/1808.03314.
[7] Gopinath, S., Ghanathe, N., Seshadri, V., & Sharma, R. (2019). Compiling KB-sized machine learning models to tiny IoT devices. Proceedings of the 40th ACM SIGPLAN Conference on Programming Language Design and Implementation.
[8] Warden, P., & Situnayake, D. (2019). TinyML: Machine Learning with TensorFlow Lite on Arduino and Ultra-Low-Power Microcontrollers.
[9] Abadi, M., Barham, P., Chen, J., Chen, Z., Davis, A., Dean, J., Devin, M., Ghemawat, S., Irving, G., Isard, M., Kudlur, M., Levenberg, J., Monga, R., Moore, S., Murray, D.G., Steiner, B., Tucker, P.A., Vasudevan, V., Warden, P., Wicke, M., Yu, Y., & Zhang, X. (2016). TensorFlow: A system for large-scale machine learning. USENIX Symposium on Operating Systems Design and Implementation.
[10] Krishnamoorthi, R. (2018). Quantizing deep convolutional networks for efficient inference: A whitepaper. ArXiv, abs/1806.08342.
[11] Howard, A.G., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., Andreetto, M., & Adam, H. (2017). MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications. ArXiv, abs/1704.04861.
[12] Sandler, M., Howard, A.G., Zhu, M., Zhmoginov, A., & Chen, L. (2018). MobileNetV2: Inverted Residuals and Linear Bottlenecks. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 4510-4520.
[13] Smith, R.W. (2007). An Overview of the Tesseract OCR Engine. Ninth International Conference on Document Analysis and Recognition (ICDAR 2007), 2, 629-633.
[14] Redmon, J., Divvala, S.K., Girshick, R.B., & Farhadi, A. (2015). You Only Look Once: Unified, Real-Time Object Detection. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 779-788.
[15] Hsu, G.J., Chen, J., & Chung, Y. (2013). Application-Oriented License Plate Recognition. IEEE Transactions on Vehicular Technology, 62, 552-561.