Indoor Semantic Segmentation using depth information
Laurent Najman
- Fonction : Auteur
- PersonId : 28
- IdHAL : laurent-najman
- ORCID : 0000-0002-6190-0235
- IdRef : 087172712
Résumé
This work addresses multi-class segmentation of indoor scenes with RGB-D inputs. While this area of research has gained much attention recently, most works still rely on hand-crafted features. In contrast, we apply a multiscale convolutional network to learn features directly from the images and the depth information. We obtain state-of-the-art on the NYU-v2 depth dataset with an accuracy of 64.5%. We illustrate the labeling of indoor scenes in videos sequences that could be processed in real-time using appropriate hardware such as an FPGA.
Format du dépôt | Notice |
---|---|
Type de dépôt | Communication dans un congrès |
Titre |
en
Indoor Semantic Segmentation using depth information
|
Résumé |
en
This work addresses multi-class segmentation of indoor scenes with RGB-D inputs. While this area of research has gained much attention recently, most works still rely on hand-crafted features. In contrast, we apply a multiscale convolutional network to learn features directly from the images and the depth information. We obtain state-of-the-art on the NYU-v2 depth dataset with an accuracy of 64.5%. We illustrate the labeling of indoor scenes in videos sequences that could be processed in real-time using appropriate hardware such as an FPGA.
|
Auteur(s) |
Camille Couprie
1
, Clément Farabet
2, 3
, Laurent Najman
3
, Yann Lecun
2
1
IFPEN -
IFP Energies nouvelles
( 300006 )
- 1-4 avenue de Bois Préau
92500 Rueil-Malmaison
- France
2
CIMS -
Courant Institute of Mathematical Sciences [New York]
( 224472 )
- 251, Mercier Street, New York, NY 10012
- États-Unis
3
LIGM -
Laboratoire d'Informatique Gaspard-Monge
( 3210 )
- Université de Paris-Est Marne-la-Vallée, Cité Descartes, Bâtiment Copernic, 5 bd Descartes, 77454 Marne-la-Vallée Cedex 2
- France
|
Vulgarisation |
Non
|
Comité de lecture |
Oui
|
Actes |
Oui
|
Invité |
Non
|
Langue du document |
Anglais
|
Date de production/écriture |
2013-03-14
|
Titre de l'ouvrage |
Proceedings of the International Conference on Learning Representations
|
Audience |
Internationale
|
Date de publication |
2013
|
Page/Identifiant |
1-8
|
Titre du congrès |
First International Conference on Learning Representations (ICLR 2013)
|
Date début congrès |
2013-05-02
|
Date fin congrès |
2013-05-04
|
Ville |
Scottsdale, AZ
|
Pays |
États-Unis
|
Commentaire |
8 pages, 3 figures
|
Domaine(s) |
|
arXiv Id | 1301.3572 |
Loading...