Abstract

Depth images have granted new possibilities to computer vision researchers across the field. A prominent task is scene understanding and segmentation on which the present work is concerned. In this paper, we present a procedure combining well known methods in a unified learning framework based on stacked classifiers; the benefits are two fold: on one hand, the system scales well to consider different types of complex features and, on the other hand, the use of stacked classifiers makes the performance of the proposed technique more accurate. The proposed method consists of a random forest using random offset features in combination with a conditional random field (CRF) acting on a simple linear iterative clustering (SLIC) superpixel segmentation. The predictions of the CRF are filtered spatially by a multi-scale decomposition before merging it with the original feature set and applying a stacked random forest which gives the final predictions. The model is tested on the renown NYU-v2 dataset and the recently available SUNRGBD dataset. The approach shows that simple multimodal features with the power of using multi-class multi-scale stacked sequential learners (MMSSL) can achieve slight better performance than state of the art methods on the same dataset. The results show an improvement of 2.3% over the base model by using MMSSL and displays that the method is effective in this problem domain.
Original languageEnglish
JournalPattern Recognition Letters
Volume80
Pages (from-to)208–215
ISSN0167-8655
DOIs
Publication statusPublished - 2016

Fingerprint

Classifiers
Merging
Computer vision
Decomposition

Keywords

  • RGB-D sematic segmentation
  • Stacked sequential learning
  • Conditional random fields
  • Random forests using random offset features

Cite this

@article{2376ce860d604538aaea446304e90a1c,
title = "Segmentation of RGB-D indoor scenes by stacking random forests and conditional random fields",
abstract = "Depth images have granted new possibilities to computer vision researchers across the field. A prominent task is scene understanding and segmentation on which the present work is concerned. In this paper, we present a procedure combining well known methods in a unified learning framework based on stacked classifiers; the benefits are two fold: on one hand, the system scales well to consider different types of complex features and, on the other hand, the use of stacked classifiers makes the performance of the proposed technique more accurate. The proposed method consists of a random forest using random offset features in combination with a conditional random field (CRF) acting on a simple linear iterative clustering (SLIC) superpixel segmentation. The predictions of the CRF are filtered spatially by a multi-scale decomposition before merging it with the original feature set and applying a stacked random forest which gives the final predictions. The model is tested on the renown NYU-v2 dataset and the recently available SUNRGBD dataset. The approach shows that simple multimodal features with the power of using multi-class multi-scale stacked sequential learners (MMSSL) can achieve slight better performance than state of the art methods on the same dataset. The results show an improvement of 2.3{\%} over the base model by using MMSSL and displays that the method is effective in this problem domain.",
keywords = "RGB-D sematic segmentation, Stacked sequential learning, Conditional random fields, Random forests using random offset features",
author = "Mikkel Th{\o}gersen and Guerrero, {Sergio Escalera} and Jordi Gonz{\`a}lez and Moeslund, {Thomas B.}",
year = "2016",
doi = "10.1016/j.patrec.2016.06.024",
language = "English",
volume = "80",
pages = "208–215",
journal = "Pattern Recognition Letters",
issn = "0167-8655",
publisher = "Elsevier",

}

Segmentation of RGB-D indoor scenes by stacking random forests and conditional random fields. / Thøgersen, Mikkel; Guerrero, Sergio Escalera; Gonzàlez, Jordi; Moeslund, Thomas B.

In: Pattern Recognition Letters, Vol. 80, 2016, p. 208–215.

Research output: Contribution to journalJournal articleResearchpeer-review

TY - JOUR

T1 - Segmentation of RGB-D indoor scenes by stacking random forests and conditional random fields

AU - Thøgersen, Mikkel

AU - Guerrero, Sergio Escalera

AU - Gonzàlez, Jordi

AU - Moeslund, Thomas B.

PY - 2016

Y1 - 2016

N2 - Depth images have granted new possibilities to computer vision researchers across the field. A prominent task is scene understanding and segmentation on which the present work is concerned. In this paper, we present a procedure combining well known methods in a unified learning framework based on stacked classifiers; the benefits are two fold: on one hand, the system scales well to consider different types of complex features and, on the other hand, the use of stacked classifiers makes the performance of the proposed technique more accurate. The proposed method consists of a random forest using random offset features in combination with a conditional random field (CRF) acting on a simple linear iterative clustering (SLIC) superpixel segmentation. The predictions of the CRF are filtered spatially by a multi-scale decomposition before merging it with the original feature set and applying a stacked random forest which gives the final predictions. The model is tested on the renown NYU-v2 dataset and the recently available SUNRGBD dataset. The approach shows that simple multimodal features with the power of using multi-class multi-scale stacked sequential learners (MMSSL) can achieve slight better performance than state of the art methods on the same dataset. The results show an improvement of 2.3% over the base model by using MMSSL and displays that the method is effective in this problem domain.

AB - Depth images have granted new possibilities to computer vision researchers across the field. A prominent task is scene understanding and segmentation on which the present work is concerned. In this paper, we present a procedure combining well known methods in a unified learning framework based on stacked classifiers; the benefits are two fold: on one hand, the system scales well to consider different types of complex features and, on the other hand, the use of stacked classifiers makes the performance of the proposed technique more accurate. The proposed method consists of a random forest using random offset features in combination with a conditional random field (CRF) acting on a simple linear iterative clustering (SLIC) superpixel segmentation. The predictions of the CRF are filtered spatially by a multi-scale decomposition before merging it with the original feature set and applying a stacked random forest which gives the final predictions. The model is tested on the renown NYU-v2 dataset and the recently available SUNRGBD dataset. The approach shows that simple multimodal features with the power of using multi-class multi-scale stacked sequential learners (MMSSL) can achieve slight better performance than state of the art methods on the same dataset. The results show an improvement of 2.3% over the base model by using MMSSL and displays that the method is effective in this problem domain.

KW - RGB-D sematic segmentation

KW - Stacked sequential learning

KW - Conditional random fields

KW - Random forests using random offset features

UR - http://www.sciencedirect.com/science/article/pii/S016786551630157X

U2 - 10.1016/j.patrec.2016.06.024

DO - 10.1016/j.patrec.2016.06.024

M3 - Journal article

VL - 80

SP - 208

EP - 215

JO - Pattern Recognition Letters

JF - Pattern Recognition Letters

SN - 0167-8655

ER -