2024 Faculty Courses School of Computing Department of Computer Science Graduate major in Artificial Intelligence
3D Computer Vision
- Academic unit or major
- Graduate major in Artificial Intelligence
- Instructor(s)
- Asako Kanezaki
- Class Format
- Lecture (HyFlex)
- Media-enhanced courses
- -
- Day of week/Period
(Classrooms) - 3-4 Tue / 3-4 Fri
- Class
- -
- Course Code
- ART.T466
- Number of credits
- 200
- Course offered
- 2024
- Offered quarter
- 4Q
- Syllabus updated
- Mar 14, 2025
- Language
- English
Syllabus
Course overview and goals
This course teaches how to process 3D data to extract higher-level information from the data. It covers the basics of data processing, geometric transformations, and linear algebra for machine learning, as well as recent cutting-edge research on deep neural networks, with hands-on exercises using programming languages such as Python.
Course description and aims
At the end of this course, students should be able to
1) acquire the basics of 3D data processing, and
2) implement their desired processing, including deep learning with, e.g., Python.
Keywords
3D Data Processing, Geometric Transformations, Object Recognition, Deep Learning, Neural Networks
Competencies
- Specialist skills
- Intercultural skills
- Communication skills
- Critical thinking skills
- Practical and/or problem-solving skills
Class flow
The lecture will cover the basics, applications, and practices of 3D data processing using lecture materials (slides) and programming exercises.
Course schedule/Objectives
| Course schedule | Objectives | |
|---|---|---|
| Class 1 | Guidance, Introduction of 3D data processing | Understand 3D sensors and data acquisition methods. |
| Class 2 | I/O and Rendering | Understand file input/output and data rendering methods. |
| Class 3 | Rotation, Translation, Scale, and Sampling | Understand geometric transformations and sampling methods for 3D data. |
| Class 4 | Normal estimation, Keypoints, and Features | Understand normal vector estimation methods, 3D keypoint detection methods, and 3D features. |
| Class 5 | Point Cloud Registration (Nearest neighbor search, ICP) | Understand k-d tree data structures and nearest neighbor search, RANSAC, and ICP algorithms. |
| Class 6 | Point Cloud Registration (Implementation) | Implement point cloud registration and apply it to 3D data. |
| Class 7 | Object classification, Pose estimation | Learn about various object recognition tasks and understand the methods. |
| Class 8 | Primitive detection, Segmentation | Learn about various object recognition tasks and understand the methods. |
| Class 9 | Fundamentals of Deep Learning | Learn methods of image classification using deep learning |
| Class 10 | PointNet, Point cloud convolution | Learn and implement point cloud processing methods using deep learning. |
| Class 11 | Latest research trends, General object pose estimation | Learn about the latest research trends in 3D data processing using deep learning and understand general object pose estimation. |
| Class 12 | RGBD, Voxel data, Mesh, Multi-view images, and Implicit functions | Learn about data formats other than 3D point clouds and understand the latest research trends using implicit functions. |
| Class 13 | Student Presentation (1) - 2-3 min. talk for each student | Each student will give an oral presentation, ask questions, and discuss others' presentations. |
| Class 14 | Student Presentation (2) - 2-3 min. talk for each student | Each student will give an oral presentation, ask questions, and discuss others' presentations. |
Study advice (preparation and review)
To improve the effectiveness of learning, students are expected to prepare for and review (including assignments) the contents of each class for approximately 60 minutes each, referring to the appropriate sections of the textbook and handouts.
Textbook(s)
None
Reference books, course materials, etc.
Materials translated into English from the above-mentioned Japanese reference book will be distributed.
Evaluation methods and criteria
Evaluation of the student's understanding of the lecture content. Grades will be based on attendance and the final presentation.
Related courses
- ART.T467 ： Fundamentals of Computer Vision
- ART.T552 ： Advanced Topics in Computer Vision
- ART.T551 ： Image and Video Recognition
Prerequisites
There are no prerequisites for this course.