Multiple View Object Cosegmentation using Appearance and Stereo Cues

Published on 2012-11-124903 Views

Adarsh Kowdle

We present an automatic approach to segment an object in calibrated images acquired from multiple viewpoints. Our system starts with a new piecewise planar layer-based stereo algorithm that estimates

ECCV 2012 - Firenze

Related categories

Object Recognition

Presentation

Multiple View Object Cosegmentation using Appearance and Stereo Cues00:00

Image 100:04

Final result using our approach00:19

Previous Work (1)00:23

Previous Work (2)00:59

Contributions01:40

Overview (1)01:56

Overview (2)02:07

Piecewise planer stereo (1)02:25

Piecewise planer stereo (2)02:31

Piecewise planer stereo (3)02:46

Piecewise planer stereo (4)03:00

Piecewise planer stereo (5)03:09

Piecewise planer stereo (6)03:18

Piecewise planer stereo (7)03:42

Piecewise planer stereo (8)04:03

Piecewise planer stereo (9)04:07

Piecewise planer stereo (10)04:26

Piecewise planer stereo (11)04:45

Piecewise planer stereo (12)05:04

Piecewise planer stereo (13)05:09

Overview (3)05:25

Region-level FG/BG labeling (1)05:30

Region-level FG/BG labeling (2)05:36

Region-level FG/BG labeling (3)05:45

Region-level FG/BG labeling (4)06:24

Region-level FG/BG labeling (5)06:32

Region-level FG/BG labeling (6)06:39

Region-level FG/BG labeling (7)06:49

Region-level FG/BG labeling (9)07:01

Region-level FG/BG labeling (10)07:06

Region-level FG/BG labeling (11)07:17

Region-level FG/BG labeling (12)07:30

Region-level FG/BG labeling (13)07:40

Overview (4)07:53

Multiview FG/BG labeling (1)07:58

Multiview FG/BG labeling (2)08:06

Multiview FG/BG labeling (3)08:32

Multiview FG/BG labeling (4)08:40

Multiview FG/BG labeling (5)08:52

Multiview FG/BG labeling (6)08:57

Overview (5)09:09

Datasets09:14

Quantitative results (1)09:29

Quantitative results (2)09:35

Quantitative results (3)09:44

Quantitative results (4)09:47

Quantitative results (5)09:57

Comparisons (1)10:00

Comparisons (2)10:35

Quantitative results (6)10:53

Quantitative results (7)11:06

Quantitative results (8)11:17

Conclusions11:30

Thank you11:50

Additional Results11:54