Fast and accurate unknown object segmentation for robotic systems
(2013) 9th International Symposium on Advances in Visual Computing, ISVC 2013 In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 8034 LNCS(PART 2). p.318-327- Abstract
Object segmentation is the first step towards more advanced robotic behaviors, as robots need to localize objects before attempting tasks such as grasping or manipulation. A robot vision system should be able to provide accurate object hypotheses in reasonably high frame rates, using images and possibly also depth data. This work proposes a fixation-based object segmentation algorithm able to cope with unknown objects, and run on a real-time robot. We show that a balanced combination of moderately accurate, when considered independently, but at the same time computationally inexpensive building modules can yield remarkable results both in terms of accuracy, but also of execution speed. We describe our algorithm and present both... (More)
Object segmentation is the first step towards more advanced robotic behaviors, as robots need to localize objects before attempting tasks such as grasping or manipulation. A robot vision system should be able to provide accurate object hypotheses in reasonably high frame rates, using images and possibly also depth data. This work proposes a fixation-based object segmentation algorithm able to cope with unknown objects, and run on a real-time robot. We show that a balanced combination of moderately accurate, when considered independently, but at the same time computationally inexpensive building modules can yield remarkable results both in terms of accuracy, but also of execution speed. We describe our algorithm and present both qualitative and quantitative experimental results that indicate significant speed-up over the state-of-the-art.
(Less)
- author
- Nalpantidis, Lazaros ; Großmann, Bjarne and Krüger, Volker LU
- publishing date
- 2013-11-29
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- keywords
- Grab Cut, Graph Cut, object segmentation, robot vision, unknown objects
- host publication
- Advances in Visual Computing : 9th International Symposium, ISVC 2013, Proceedings - 9th International Symposium, ISVC 2013, Proceedings
- series title
- Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
- volume
- 8034 LNCS
- issue
- PART 2
- pages
- 10 pages
- conference name
- 9th International Symposium on Advances in Visual Computing, ISVC 2013
- conference location
- Rethymnon, Crete, Greece
- conference dates
- 2013-07-29 - 2013-07-31
- external identifiers
-
- scopus:84888273432
- ISSN
- 0302-9743
- 1611-3349
- ISBN
- 9783642419386
- DOI
- 10.1007/978-3-642-41939-3_31
- language
- English
- LU publication?
- no
- id
- 3a17a3c5-1ae6-47e5-8af4-55d09576a520
- date added to LUP
- 2019-06-28 09:19:42
- date last changed
- 2024-08-07 00:46:58
@inproceedings{3a17a3c5-1ae6-47e5-8af4-55d09576a520, abstract = {{<p>Object segmentation is the first step towards more advanced robotic behaviors, as robots need to localize objects before attempting tasks such as grasping or manipulation. A robot vision system should be able to provide accurate object hypotheses in reasonably high frame rates, using images and possibly also depth data. This work proposes a fixation-based object segmentation algorithm able to cope with unknown objects, and run on a real-time robot. We show that a balanced combination of moderately accurate, when considered independently, but at the same time computationally inexpensive building modules can yield remarkable results both in terms of accuracy, but also of execution speed. We describe our algorithm and present both qualitative and quantitative experimental results that indicate significant speed-up over the state-of-the-art.</p>}}, author = {{Nalpantidis, Lazaros and Großmann, Bjarne and Krüger, Volker}}, booktitle = {{Advances in Visual Computing : 9th International Symposium, ISVC 2013, Proceedings}}, isbn = {{9783642419386}}, issn = {{0302-9743}}, keywords = {{Grab Cut; Graph Cut; object segmentation; robot vision; unknown objects}}, language = {{eng}}, month = {{11}}, number = {{PART 2}}, pages = {{318--327}}, series = {{Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)}}, title = {{Fast and accurate unknown object segmentation for robotic systems}}, url = {{http://dx.doi.org/10.1007/978-3-642-41939-3_31}}, doi = {{10.1007/978-3-642-41939-3_31}}, volume = {{8034 LNCS}}, year = {{2013}}, }