2010 IEEE International Conference on Image Processing 2010
DOI: 10.1109/icip.2010.5651643
|View full text |Cite
|
Sign up to set email alerts
|

Cue-independent extending inverse kinematics for robust pose estimation in 3D point clouds

Abstract: While monocular gesture recognition slowly reaches maturity, the inclusion of 3D gestures remains a challenge. In order to enable robust and versatile depth-enabled gestures, a depth-image based tracking approach is developed. Using a model-based annealing particle filter approach, the pose of a single subject is retrieved and tracked over longer image and motion sequences. Other than many previous depth-image based systems, full body tracking is performed. The system is independent from specific camera types … Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1

Citation Types

0
4
0

Year Published

2011
2011
2013
2013

Publication Types

Select...
4
2
1

Relationship

1
6

Authors

Journals

citations
Cited by 7 publications
(4 citation statements)
references
References 10 publications
0
4
0
Order By: Relevance
“…To evaluate the potential for real-time tracking applications, the proposed observation likelihood function has been implemented in CUDA and used as a weighting function within an APF based upper body pose tracker (derived from [10]). It was found that the implementation evaluated 300 sample poses with 288 points each against an average of 550 observed points in less than 25 ms.…”
Section: Performance Using Cudamentioning
confidence: 99%
“…To evaluate the potential for real-time tracking applications, the proposed observation likelihood function has been implemented in CUDA and used as a weighting function within an APF based upper body pose tracker (derived from [10]). It was found that the implementation evaluated 300 sample poses with 288 points each against an average of 550 observed points in less than 25 ms.…”
Section: Performance Using Cudamentioning
confidence: 99%
“…Our proposal achieves 68 f ps on a regular CPU, which is about 4−5× faster than the cited method. Moreover, some of the proposals in Table 10.2 use GPU implementations [LKAR10,GPKT10a], which should speed their performance up.…”
Section: Handbox Vs Reference Methodsmentioning
confidence: 99%
“…As the roas only one degree of free- Figure 3.10: System diagram of the entire pipeline in [132]. The authors in [56] employed a model-based annealing particle filter (APF) approach to retrieve and track the pose of a single subject in video sequences. GPU is used to accelerate the system with 300 points to run at 2.5 fps.…”
Section: State-of-the-art Of Markerless Motion Capture Based On the Rmentioning
confidence: 99%
“…Different poses with the matched body prototype overlaid results in[56].The authors in[5] proposed a data-driven hybrid strategy that combines local optimization with global retrieval techniques to estimation full body pose based on a single depth image stream. The overview of system is shown inFigure 3.13.…”
mentioning
confidence: 99%