Directions in Shape Analysis towards Functionality

SIGGRAPH Asia 2016 Course

Abstract 

The majority of man-made objects are designed to serve a certain function, and this is often reflected by the geometry of the objects, or the way that they are used or organized in an environment. In recent years, many efforts in shape analysis have developed methods that extract high-level structural and semantic information from geometric shapes and scenes, especially involving man-made objects. One can argue that the ultimate goal of some of these works is to understand the functionality of the objects. Moreover, there have also been works that explicitly model and incorporate functionality into the processing of shapes and scenes. Thus, functionality has been receiving increasingly more attention in shape analysis and geometric modeling, either directly or indirectly, since functionality considerations can aid in applications such as semantic classification, shape editing and synthesis, as well as product design, development, and fabrication.

In this course, we discuss recent developments that incorporate functionality aspects into the analysis of 3D shapes and scenes, to provide a summary of the state-of-the-art in this area, including a discussion of key ideas and literature works. More specifically, we first discuss approaches that are precursors in this front, such as structure-aware and data-driven methods that learn relationships between shape parts or objects in scenes. Next, we cover works that more explicitly model the functionality of shapes and scenes, such as agent- and interaction-based methods. The course is structured in the form of talks given by four different speakers, aided by electronic slides that include notes for subsequent consultation.


Schedule

Topic

Instructor

Introduction (15 mins)

Introduction to the topic of functionality analysis, description of what is covered in the course, and rationale and motivation for the course. 

Oliver van Kaick

Structure-aware and data-driven works oriented to functionality analysis (45 mins)

- Co-segmentation and co-hierarchies of shape collections
- Labeling methods for functionality recognition
- Meta-representation and co-constrained handles 
- Symmetry analysis for structure discovery

Oliver van Kaick

Form and function: from structure analysis to functional manipulation and modeling (45 mins)

- Recurring part arrangements in shape collections 
- Structure-preserving shape manipulation
- Ergonomics-inspired geometric exploration 

Youyi Zheng

Break (15 mins)

Human-centric analysis for understanding interactions (45 mins)

- Human-centric analysis for understanding interactions
- Inferring action maps in 3D environments
- Activity-centric 3D scene synthesis
- Learning to generate 3D interactions from language

Manolis Savva

Functionality analysis based on general object-object interaction (45 mins)

- Geometric representation of object-object interaction
- Interaction-based complex scene synthesis
- Context-based functionality descriptor
- Functionality model based on co-analysis of interactions

Ruizhen Hu

Conclusion (15 mins)

Summary of the state-of-the-art, and discussion of future challenges and possible future work.

Ruizhen Hu


Instructor Bios

Ruizhen HuShenzhen University

She is an Assistant Professor at Shenzhen University, China. She received her Ph.D. from the Department of Mathematics, Zhejiang University. Before that, she spent two years visiting Simon Fraser University, Canada. Ruizhen’s research interests are in shape analysis, geometry processing and fabrication.





Oliver van KaickCarleton University

He is an Assistant Professor at Carleton University, Ottawa, Canada. He received a Ph.D. from the School of Computing Science at Simon Fraser University (SFU). Oliver was then a postdoctoral researcher at SFU and Tel Aviv University. Oliver’s research is concentrated in shape analysis and geometric modeling. 





Youyi ZhengShanghaiTech University

He is currently an Assistant Professor at the School of Information Science and Technology, ShanghaiTech University. He obtained his PhD from the Department of Computer Science and Engineering at Hong Kong University of Science & Technology, and his M.Sc. and B.Sc. degrees from the Department of Mathematics, Zhejiang University. His research interests include geometric modeling, imaging, and human-computer interaction.





Manolis SavvaPrinceton University

He is a postdoctoral scholar the Princeton Graphics Group. He obtained his PhD at the Stanford Graphics Lab advised by Pat Hanrahan. His research focuses on human-centric 3D scene analysis and scene generation. He has also worked in data visualization, grounding of natural language to 3D content, and more recently in establishing the ShapeNet large-scale 3D model dataset.



Copyright © 2016-2018 Visual Computing Research Center