Sign inSign up
Metric Regression Forests for Human Pose Estimation thumbnail
Pause
Mute
Subtitles
Playback speed
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Full screen

Metric Regression Forests for Human Pose Estimation

Published on Apr 03, 20142548 Views

Gerard Pons-Moll

We present a new method for inferring dense data to model correspondences, focusing on the application of human pose estimation from depth images. Recent work proposed the use of regression forests

BMVC 2013 - Bristol

Related categories

Computer VisionMachine Learning

Chapter list

Metric Regression Forests for Human Pose Estimation00:00
Human Pose Estimation - 100:10
Human Pose Estimation - 200:22
From Classification to Regression - 100:34
From Classification to Regression - 201:04
From Classification to Regression - 301:30
From Classification to Regression - 401:34
From Classification to Regression - 501:42
From Classification to Regression - 601:54
Inferring Correspondences - 102:08
Inferring Correspondences - 202:20
Inferring Correspondences - 302:22
Inferring Correspondences - 402:34
Inferring Correspondences - 502:37
Parts Classification Objective02:40
Retrofitting for Regression - 103:44
Retrofitting for Regression - 203:59
Retrofitting for Regression - 304:02
Retrofitting for Regression - 404:11
Current Limitations - 104:27
Current Limitations - 204:42
Current Limitations - 304:45
Current Limitations - 404:54
Current Limitations - 505:13
How much better can we do?05:30
Metric Space07:01
Training with a Differential Entropy Objective07:47
Kernel Density Estimation - 108:37
Kernel Density Estimation - 208:49
Kernel Density Estimation - 309:09
How to Compute it? - 110:18
How to Compute it? - 210:23
How to Compute it? - 310:24
How to Compute it? - 410:34
Problems - 110:50
Problems - 211:12
Our approach: Discretize and Cache Kernel11:30
Approximate differential entropy - 111:59
Approximate differential entropy - 212:13
Approximate differential entropy - 312:38
Approximate differential entropy - 412:42
Entropy is computed as - 112:53
Entropy is computed as - 213:10
Results13:40
Optimize the Pose - 114:33
Optimize the Pose - 214:43
Optimize the Pose - 314:47
Pose Accuracy14:58
Improvement after ICP15:15
Where are we?16:05
Conclusions16:36
Thank you!17:24