Dynamical Pose Filtering for Mixtures of Gaussian Processes thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Dynamical Pose Filtering for Mixtures of Gaussian Processes

Published on Oct 09, 20122987 Views

Related categories

Chapter list

Discriminative Pose Tracking with Mixtures of Gaussian Processes00:00
Discriminative Pose Estimation00:13
Challenges01:01
Contributions01:50
Mixture of Experts Model02:06
Gaussian Processes03:02
Applying Gaussian Processes into a Mixture of Experts Model04:44
Optimising the Expert Indicators05:55
Obtaining a Pose Estimate07:53
Incorporating a Dynamics Constraint08:43
Dynamical Pose Filtering09:32
Second Order Dynamics10:12
Inferring the Optimal Observation Sequence (1)11:23
Inferring the Optimal Observation Sequence (2)11:34
Inferring the Optimal Observation Sequence (3)11:37
Inferring the Optimal Observation Sequence (4)11:58
Inferring the Optimal Observation Sequence (5)12:59
Results on Human Pose Tracking13:08
Quantitive Results: Appearance Model14:28
Quantitive Results: Dynamics Model 014:52
Questions15:50