Indoor Semantic Segmentation using depth information - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2013

Indoor Semantic Segmentation using depth information

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.

Dates et versions

hal-00805105 , version 1 (27-03-2013)

Identifiants

Citer

Camille Couprie, Clément Farabet, Laurent Najman, Yann Lecun. Indoor Semantic Segmentation using depth information. First International Conference on Learning Representations (ICLR 2013), May 2013, Scottsdale, AZ, United States. pp.1-8. ⟨hal-00805105⟩
1094 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More