|
-
June 3. The slides and the lecture notes on "Surface Reconstruction" are available here. Please note that the notes mainly contain three research papers, and you're highly encouraged (but not obliged) to read them.
-
May 20. The slides and lecture notes on "Multi-view stereo (MVS)" are available here.
-
All news ...
|
About
Photogrammetry and 3D Computer Vision (i.e., 3DV) aim at recovering the structure of real-world objects/scenes from images. This course is about the theories, methodologies, and techniques of 3D computer vision for the built environment. In the term of this course, you will learn the basic knowledge and algorithms in 3D computer vision through a series of lectures, reading materials, lab exercises, and group assignments.
Contents
The topics of this course cover the whole pipeline of reconstructing 3D models from images:
-
Cameras models: how a point from the real world gets projected onto the image plane and how to recover the camera parameters from a set of observations;
-
Epipolar geometry: the geometric relations between 3D points and their images points; the constraints between the image points;
-
Image matching: define and match image features (SIFT) to establish correspondences between images;
-
Structure from motion: recover/refine geometry and camera parameters from a set of images;
-
Multi-view stereo and learning-based approaches for recovering dense geometry (e.g., point clouds) from images;
-
Surface reconstruction: obtain 3D surface models of real-world objects from point clouds;
-
Semantic segmentation: classify/segment the reconstructed 3D point; into semantically meaningful object classes.
Goals
After finishing this course, you will have gained knowledge of 3D computer vision techniques and the skills to apply them for recovering 3D geometry of real-world objects from images. Specifically, you will be able to:
-
apply linear algebra knowledge to implement basic 3D computer vision algorithms;
-
explain the main concepts in 3D computer vision (e.g., camera models, epipolar constraints, essential matrix, fundamental matrix, image matching, triangulation, structure from motion, bundle adjustment, multi-view stereo, and surface reconstruction);
-
explain the principles of the state-of-the-art 3D computer vision pipelines for 3D dense reconstruction from images;
-
choose appropriate methods for reconstructing smooth surfaces and piecewise planar objects; explain and evaluate these methods;
-
propose and implement solutions for reconstructing real-world buildings from images;
Assessment
The assessment of this course consists of 3 group assignments and the final exam. The final grade will be based on the contribution in doing the assignments and the final exam, breaking down to:
-
group assignments (40%). All assignments have equal weight in the final grade. It is possible to resubmit your work after incorporating the feedback/suggestions received from the teachers. However, the evaluation of an assignment is mainly based on the first submission. Students who have improved their work may receive a slightly higher grade depending on the significance of the improvement (but no more than 0.5).
-
final exam (60%). The final exam consists of multiple-choice questions and open questions. Example questions will be given two weeks before the exam.
Both assignments and the final exam have to meet the minimum requirement (i.e., 5.5) to pass the course. A total of 6.0 or above is necessary to pass the course.
Prerequisites
- Proficiency in C++ programming.
All assignments will be in C++ (because of its good performance). If you have a lot of programming experience but in a different language (e.g., Python/Matlab/Javascript) you will probably be fine. For those who are not familiar with C++, please check the Programming resources and tools.
-
Basic linear algebra and calculus.
You should be comfortable understanding matrix-vector operations and notation.
Course logistics
-
Lectures: on campus. Check here for an overview of all the lectures.
-
Slides and lecture notes: will be posted here shortly after each lecture.
-
Office hours: Wednesday (13:45-15:45), Friday (10:45-12:45, 13:45-15:45). Course schedule and calendar
-
Contact: we will also use Discord channels for announcements and all course-related questions. We highly recommend raising questions on the Discord text channel, so others can also benefit from the Q&A. For external inquiries, emergencies, or personal matters that you don't wish to put in a private post, you can email the teacher(s) or schedule a private meeting (to make an appointment, please indicate a few time slots during the office hours and include a Zoom link).
Lecturers and teaching assistant
Copyright @Liangliang Nan. 2021
