Hauptseminar: Motion Analysis, Learning and Tracking
Nassir Navab, Diana Mateus, Loren Schwarz
Type: Advanced Seminar Course
Module IN8901
Type: Master Seminar
Module IN2107
Type: Seminar
Module IN0014
SWS: 2+0
ECTS: 4 Credits
Time/Location: See table below
Course Language: English
Content
The seminar's participants will get a comprehensive insight into the latest developments in the domain of computer-vision-based motion analysis. The focus of this seminar is on human motion, although topics on general object motion are also available. Being able to reconstruct human poses given only limited observations, such as 2D images or wearable sensor data, is a challenging task. However, various potential fields of application, e.g. video surveillance or medical motion analysis, motivate research in this direction.
The focus of this seminar is on the following topics:
- Acquisition of human motion data using motion capture
- Representation of human motion in low-dimensional subspaces
- Analysis of human motion characteristics for medical purposes
- Machine learning approaches for human motion tracking
- Human activity recognition techniques
The talks will be based on the latest proceedings of medical and computer science conferences, such as the International Conference on Computer Vision (ICCV), the International Conference on Computer Vision and Pattern Recognition (CVPR) and others.
Registration
Interested students, please take a look at the papers listed below and select one you would like to base your seminar talk on. Then please send an email to
Loren Schwarz (), including your paper choice, your name, your birthday, program of study and Matrikelnummer. You can also propose other topics or papers related to motion analysis.
Requirements
In order to successfully complete the seminar, participants are required to (1)
present the selected paper to the other participants and (2) submit a written
handout. Each presentation (about 30-45min) is followed by a discussion, and everyone is encouraged to actively participate. The handout (about 10 pages) must include all references used and must be written completely in your own words and not by copy and paste. Both handout and presentation have to be done in
English.
Presentation Topics
Human Motion Analysis and Tracking
- (taken) Fossati A., Salzmann M. and Fua P.: Observable subspaces for 3D human motion recovery. Computer Vision and Pattern Recognition (CVPR), 2009
- Huang P., Hilton A. and Starck J.: Human motion synthesis from 3D video. Computer Vision and Pattern Recognition (CVPR), 2009
- Urtasun R. and Darrell T.: Sparse probabilistic regression for activity-independent human pose inference. Computer Vision and Pattern Recognition (CVPR), 2008
- (taken) Ali S. and Shah M.: Floor fields for tracking in high density crowd scenes. European Conference on Computer Vision (ECCV), 2008
- Gupta A., Chen T., Kimber D. and Davis L.: Context and observation driven latent variable models for human pose estimation. Computer Vision and Pattern Recognition (CVPR), 2008
- Wang J., Fleet D. and Hertzmann A.: Gaussian process dynamical models for human motion. Pattern Analysis and Machine Intelligence (PAMI), 2008
- Hou S., Galata A., Caillette F and Thacker N.: Real-time body tracking using a Gaussian process latent variable model. International Conference on Computer Vision (ICCV), 2007
- (taken) Vlasic D., Adelsberger R., Vannucci G. and Barnwell J.: Practical motion capture in everyday surroundings. ACM Transactions on Graphics (TOG), 2007
- Kanaujia A., Sminchisescu C. and Metaxas D.: Spectral latent variable models for perceptual inference. International Conference on Computer Vision (ICCV), 2007
- Jenkins O., Gonzalez G and Loper M.: Dynamical motion vocabularies for kinematic tracking and activity recognition. Computer Vision and Pattern Recognition (CVPR), 2006
- (taken) Lee C. and Elgammal A.: Human motion synthesis by motion manifold learning and motion primitive segmentation. Articulated Motion and Deformable Objects (AMDO), 2006
- Chai J. and Hodgins J.: Performance animation from low-dimensional control signals. Computer Graphics and Interactive Techniques (SIGGRAPH), 2005
Human Activity Recognition
- (taken) Morris B. and Trivedi M.: Learning trajectory patterns by clustering: experimental studies and comparative evaluation. Computer Vision and Pattern Recognition (CVPR), 2009
- (taken) Tung T. and Matsuyama T.: Topology dictionary with Markov model for 3D video content-based skimming and description. Computer Vision and Pattern Recognition (CVPR), 2009
- (taken) Weinland D., Ronfard R. and Boyer E.: Free viewpoint action recognition using motion history volumes. Computer Vision and Image Understanding (CVIU), 2006
- Minnen D., Starner T., Essa I. and Isbell C.: Discovering characteristic actions from on-body sensor data. Symposium on Wearable Computers, 2006
Object Motion and Tracking
- (taken) Gosch C., Fundana K., Heyden A. and Schnörr C.: View point tracking of rigid objects based on shape sub-manifolds. Computer Vision and Pattern Recognition (CVPR), 2009
Schedule
The seminar consists of an introductory meeting and four sessions. Each of the four sessions has two student presentations. All seminar participants are expected to be present at every session and to actively participate in the discussion.