三維電腦視覺
3D Computer Vision
| 節 | 週一 |
|---|---|
5 13:20–14:10 | 三維電腦視覺 CM216 3 節連堂 |
6 14:20–15:10 | |
7 15:30–16:20 |
* 根據陽明交大上課時間表所列
This course will cover important, fundamental topics in computer graphics and computer vision that are related to 3D data and geometric models, including: 1) Basic 2D CV concepts and techniques such as filtering and image features such as SIFT. 2) Various 3D-to-2D projection models such as classic pinhole camera and equirectangular (panoramic) projection. 3) Stereo/multi-view geometry: Epipolar geometry and beyond. 4) Optimization, especially topics related to visual computing. 5) Point cloud reconstruction (e.g., triangulation), registration, and processing. 6) Latest, state-of-art research in visual computing, including but not limited to Deep Learning/AI. The goal of this course is to equip students with solid understanding and hand-on experiences in visual computing (computer vision and computer graphics), especially knowledge/skill about processing 3D data and geometric models. We are in the advent of "native 3D computing" - all visual data that were previously 2D - pictures, photos, videos, etc., are increasing becoming 3D natively thanks to the advances in modern optical hardware. Students and practitioners are best equipped with knowledge and skills to understand, process, and leverage such data. Programming projects will be done in C/C++ mostly. We will implement some fun (and a bit challenging) projects about multi-view geometry and projections, 3D point cloud registration, triangulation, and geometric processing. There will be a mid-term exam and a final exam. Exams are "open-computer", which mean that students will bring their laptops (with internet connections) and do the exams.
基本C/C++程式能力 - 本課程以C++編程為主. 基本數學觀念(線性代數,初淺的微積分,離散數學等) ---- Basic C/C++ programming skills. Basic mathematics such as basic Linear Algebra, Calculous, and Discrete Mathematics.
All materials are in English (slides, homeworks, assignments, exams). Lectures are given in Chinese.
Several homeworks. Several programming assignments. A mid-term exam and a final exam. Exams are "open-computer", which mean that students will bring their laptops (with internet connections) and do the exams.
| 週次 | 主題 |
|---|---|
| 第 1 週 | Course introduction and overview. |
| 第 2 週 | Mathematics basics and transformations. |
| 第 3 週 | 2D computer vision I: image processing and filters. |
| 第 4 週 | 2D computer vision II: image feature descriptors and matching. |
| 第 5 週 | 3D-to-2D projection models I: pinhole camera models and perspective projections. |
| 第 6 週 | 3D-to-2D projection models II: 360' panoramas and equirectangular projections. |
| 第 7 週 | SfM I: introduction to two-view ("epipolar") geometry. |
| 第 8 週 | Mid-term exam. |
| 第 9 週 | SfM II: stereo vision and depth estimation. |
| 第 10 週 | SfM III: from epipolar geometry to structure-from motion (SfM) and SLAM. |
| 第 11 週 | Geometric processing I: introduction to point clouds and acquisition methods. |
| 第 12 週 | Geometric processing II: the "point cloud-to-mesh" geometric processing pipeline. |
| 第 13 週 | Geometric processing III: 3D geometric models - mesh data structures and surface "curvature" analysis. |
| 第 14 週 | Introduction to image-based rendering and light fields. |
| 第 15 週 | 3D deep learning models overview. |
| 第 16 週 | Final exam. |
| 第 17 週 | TBA. |
| 第 18 週 | TBA. |
Richard Hartley and Andrew_Zisserman: Multiple_View_Geometry_in_Computer_Vision. Richard Szeliski: Computer Vision Algorithms and Applications 2nd Stanford Computer Vision course (by Prof. Fei-Fei Li and others), chapter 5-10. Computer Graphics course, CMU. http://15462.courses.cs.cmu.edu/fall2019/
- 地點
- 奇美樓517.
- 時間
- Email communications are welcome all the time. Face-to-face office hours are available by request.
- 聯絡方式
- pengchihan@nctu.edu.tw