Proposal for ICIG 2017 Special Session on Representation,
Analysis and Applications of Large-Scale 3D Multimedia Data

Organizers:
Dr. Sicheng Zhao (Tsinghua University)
Prof. Rongrong Ji (Xiamen University)

Motivation

3D information has been widely applied in various domains, such as virtual reality, 3D graphics, medical industry, and architecture design, which has produced explosively growing 3D multimedia data, leading to the requirement for large-scale 3D data representation and analysis. The traditional 3D feature representation and learning methods may not work well for large-scale data, since the computational cost in such cases is much larger. How to design efficient and effective 3D feature representation and learning techniques to deal with big data is desirable and meaningful. Recently, the view-based 3D analysis has attracted much research attention, due to its flexibility in representing 3D objects by multiple views, without the requirement for the complex 3D model information.

Although promising results have been achieved by existing view-based paradigm, there still remain some important problems unsolved. For example, extracting discriminative 3D features is also not well investigated, how to employ the popular deep learning methods into 3D analysis is still not clear, and how to perform 3D hashing for efficient retrieval is also an open issue. This special session will focus on the most recent progress on large-scale 3D multimedia representation, analysis and applications, such as 3D retrieval, recognition, classification, detection, and tracking, etc. The primary objective of this special session fosters focused attention on the latest research progress in this interesting area.

This special session seeks for original contributions reporting the most recent progress on 3D multimedia representation, analysis and applications. It targets a mixed audience of researchers and product developers from several communities, i.e., computer vision, multimedia, and machine learning, etc.

The topics of interest include, but are not limited to:

3D representation

•  3D data acquisition
•  3D feature extraction
•  Multi-view registration and calibration
•  3D data compression
•  3D reconstruction

3D analysis (retrieval, recognition, classification, detection, and tracking, etc.)

•  View-based 3D analysis
•  Multi-instance learning for 3D analysis
•  3D hashing
•  Deep learning for 3D data
•  3D shape analysis and morphology
•  Object tracking with RGB-D sensors
•  3D analysis on mobile devices
•  3D detection for autonomous driving
•  Point cloud for 3D analysis

3D applications

•  Medical applications, such as telemedicine
•  Industrial design, such as computer-aided manufacturing
•  Virtual reality, such as human-computer interaction and 3D synthesis

<<Back