基本素養 Basic Literacy

畢業生應具備科技人文素養及資訊工程倫理之精神
graduates should equip with both the attitude of technological/cultural literacy and the consciousness of information engineering ethics
畢業生應具備良好溝通技巧及國際觀
graduates should equip with appropriate communication skill and global view

核心能力 Competence

畢業生應具備基礎專業數學及資訊理論知識之基本能力
graduates should equip with the basic capability of the fundamental of professional mathematics and theoretical knowledge in informatics
畢業生應具備資訊理論推導及實驗設計、數據分析歸納之能力
graduates should equip with the capability of information theory derivation、experiment design and experimental data analysis/induction
畢業生應具備培養學習興趣及持續學習之能力
graduates should equip with the capability of learning interest development and continuous learning
畢業生應具備獨立、創新思維及發掘、分析、解決資訊相關問題之能力
graduates should equip with the capability to think creatively and independently and to explore, analyze, and solve information-related problems
畢業生應具備資訊系統設計、測試及驗證之能力
graduates should equip with the information system ability in designing and verification
畢業生應具備資訊系統整合之能力
graduates should equip with the capability of system integration
畢業生應具備負責之工作態度及有效團隊合作之能力
graduates should equip with a responsible attitude in working and the capability of effective team-work collaboration
具備有效溝通表達的專業語文能力

課程概述 Course Description

這門課上半學期授課將以老師的講義配合OpenCV教科書所提供較實用的程式功能論述為主,下半學期將以老師深度學習講義為主來授課。 課程內容將以常會用到的影像處理、電腦視覺及人工智慧-深度學習的理論基礎及技術為主,並學習及實作OpenCV的功能與實務應用。為了讓學生了解深度學習的原理是如何結合人工智慧及電腦視覺發展而來的相互關係,課程中會傳授影像處理中最重要的 convolution 的基本原理與實際應用,並會讓學生了解如何用神經網路來實現並強化 convolution 的功能以完善 Convolutional Neural Networks (CNN) 的架構 (topology)。講解的方法將融入老師20幾年的產學合作經驗(臉部偵測、辨識及表情分析、雲端智慧型監控服務、自動光學檢測、智慧製造、智慧型機器手臂控制、自走車)及近年來在臨床醫學影像與精準運動分析經驗來舉實際的範例解釋基礎原理,透過這些理論來解決實際的問題;技術內容包括影像處理、即時偵測、追蹤及辨識模組的設計、3D立體視覺、增擬實境及深度學習等等。 本課程期待培養學生於影像處理、電腦視覺及深度學習領域技術設計及整合實作的能力,透過務實的作業實例來培育學生具備研發思考、程式設計及解決現存問題的能力,藉由分組計劃的實作來培養學生具備發現問題、解決問題及團隊分工合作的能力與精神,並可把所學的理論基礎應用到工業界、臨床醫學影像處理及精準運動的實務面。
This class will focus on the teacher's lecture notes and the more practical functional arguments (function calls) provided by the OpenCV textbook for the first half of the semester about OpenCV. The teacher's lecture notes will be used for the second half of the semester about deep learning. The content of the course will focus on the theoretical basis and practical techniques of image processing, computer vision and artificial intelligence – deep learning. OpenCV functions and practical applications will also be learnt. In order to let students understand how the principles of depth learning combine artificial neural networks (ANNs) and the interrelationships of computer vision, the most important principles and practical applications of image processing – convolution will be taught. Therefore, students will be able to understand how to implement and enhance the functionality of the convolution in order to perfect the architecture of Convolutional Neural Networks (CNN). In addition, the lectures will incorporate teacher’s more than 20 years of experience in industrial cooperation (face detection, recognition and expression analysis, intelligent video surveillance as a service, automated optical inspection (AOI), intelligent manufacturing, visual-guided robot arm control, and automatic guided vehicle (AGV)) and recent practical examples of clinical imaging and AI Sports to explain the fundamentals and solve practical problems through these theories, including image processing, computer vision (real-time detection, tracking and recognition, 3D stereo or visualization, augmented reality (AR)) and deep learning. This course looks forward to developing students' ability to design and integrate practical skills in image processing, computer vision and depth learning, and to develop students' R&D thinking, programming design, and practical consideration through practical practice in order to have the ability to solve existing problems. Through the practical work of group project, students' ability and spirit of discovering problems, solving problems and team work-sharing can be nurtured. They can also apply the theoretical foundation to industry, clinical imaging and precision sports analytics.

課程學習目標 Course Objectives

  • 具備基本的影像處理、電腦視覺及深度學習理論基礎;
  • 融匯貫通深度學習的原理是如何結合人工智慧、電腦視覺及影像處理-卷積發展而來的;
  • 具備以影像處理、電腦視覺及深度學習的技術來解決即時偵測、追蹤、辨識及增擬實境問題
  • 學習OpenCV的功能原理及實務應用。
  • 課程進度 Progress Description

    進度說明 Progress Description
    1School doesn't begin yet
    2Holiday
    3Image processing 1 - Convolution: High-Pass (edge) and low-pass (smooth) filters, morphology operation, and CCLabeling (connected-component labeling).
    4Image processing 1 - Convolution: High-Pass (edge) and low-pass (smooth) filters, morphology operation, and CCLabeling (connected-component labeling).
    5Holiday
    6Image processing 2 - Image transforms: Hough transformation, geometric trans., FT (Fourier transform), II (integral image) and histogram equalization.
    7Image proc. 3 – (Texture) Histograms and matching: Stochastic and probability.
    8Image processing 4 – (Shape) Contours: Data structure - Linked list.
    9Sensor - Camera models and calibration: Optimization process and AR (Augmented reality).
    10Sensor - Projection and 3D vision: Geometric transformations between 2D and 3D; and 1) Stereo, 2) ToF (Time-Of-Flight, Kinect 2, SoftKinetic), and 3) Structured light (Kinect 1, DLP projector).
    11Image parts and segmentation: Background subtraction/modeling - Real-time motion detection using GMM.
    12Tracking and motion: Optical flow (feature tracking), mean-shift, Camshift tracking, and condensation algorithm (particle filter).
    13Deep Learning: From Computer Vision and BPNN (back propagation neural networks) to deep learning.Introduction to Machine Learning: 1) SIFT (and brief HOG): Feature extraction, 2) AdaBoost: Face detection, 3) PCA, LDA: Linear classification for face recognition, 4) SVM: Non-linear classification
    14Deep learning: Network Design, and Training and Test Processes of Deep Learning
    15Deep learning: Network Design, and Training and Test Processes of Deep Learning
    16Exam
    17Deep learning: Network Optimization Factors of Deep Learning
    18Deep learning: Network Optimization Factors of Deep Learning. Week 19: Final Project (Group Project)– Deep Learning
     以上每週進度教師可依上課情況做適度調整。The schedule may be subject to change.

    課程是否與永續發展目標相關調查
    Survey of the conntent relevant to SDGs

    本課程與SDGs相關項目如下:
    This course is relevant to these items of SDGs as following:
    • 教育品質 (Quality Education)
    • 就業與經濟成長 (Decent work and Economic growth)
    • 工業、創新與基礎建設 (Industry Innovation and infrastructure)

    有關課程其他調查 Other Surveys of Courses

    1.本課程是否規劃業界教師參與教學或演講? 否
    Is there any industry specialist invited in this course? How many times? No
    2.本課程是否規劃含校外實習(並非參訪)? 否
    Are there any internships involved in the course? How many hours? No
    3.本課程是否可歸認為學術倫理課程? 否
    Is this course recognized as an academic ethics course? In the course how many hours are regarding academic ethics topics? No
    4.本課程是否屬進入社區實踐課程? 否
    Is this course recognized as a Community engagement and Service learning course? Which community will be engaged? No