en-es
en
0.25
0.5
0.75
1.25
1.5
1.75
2
Multi-view Body Part Recognition with Random Forests
Published on 2014-04-033087 Views
This paper addresses the problem of human pose estimation, given images taken from multiple dynamic but calibrated cameras. We consider solving this task using a part-based model and focus on the pa
Related categories
Presentation
Multi-view Body Part Recognition with Random Forests00:00
Problem 00:09
The Typical 3D Pose Data Set00:17
Our New 3D Pose Data Set00:29
2D & 3D Pose Estimation using Pictorial Structures/Part-based Models01:01
Pictorial Structures & Part-based Models - 101:27
Pictorial Structures & Part-based Models - 202:09
Part Appearance Model02:41
2D Part Appearance Model02:49
Body Part Classification as 2D Appearance Model03:01
Decision Tree for Pixel Classification03:35
Random Forest04:18
Depth of trees and number of trees04:23
2D Pose Estimation Demo Movie04:43
3D Part Appearance Model - 105:37
3D Part Appearance Model - 205:59
The Problem of Symmetric Body Parts - 106:33
The Problem of Symmetric Body Parts - 206:42
The Problem of Symmetric Body Parts - 306:52
Aggregating Scores Across Views07:17
Naive Multi-view Pose Estimation08:03
The Problem of Symmetric Body Parts - 408:27
Handle Left-Right Correspondences with a Latent Variable09:01
Multi-view Inference09:34
3D Part Appearance Model10:15
Multi-view Pose Estimation10:42
Conclusions - 111:29
Conclusions - 211:42
Conclusions - 311:50
Thank you!12:07