TUTCRIS - Tampereen teknillinen yliopisto

TUTCRIS

Learning graph affinities for spectral graph-based salient object detection

Tutkimustuotosvertaisarvioitu

Standard

Learning graph affinities for spectral graph-based salient object detection. / Aytekin, Caglar; Iosifidis, Alexandros; Kiranyaz, Serkan; Gabbouj, Moncef.

julkaisussa: Pattern Recognition, Vuosikerta 64, 04.2017, s. 159-167.

Tutkimustuotosvertaisarvioitu

Harvard

APA

Vancouver

Author

Aytekin, Caglar ; Iosifidis, Alexandros ; Kiranyaz, Serkan ; Gabbouj, Moncef. / Learning graph affinities for spectral graph-based salient object detection. Julkaisussa: Pattern Recognition. 2017 ; Vuosikerta 64. Sivut 159-167.

Bibtex - Lataa

@article{a52061a99dd84129a8649150de80f4ba,
title = "Learning graph affinities for spectral graph-based salient object detection",
abstract = "In this paper, we propose a novel method for learning graph affinities for salient object detection. First, we assume that a graph representation of an image is given with a predetermined connectivity rule and representative features for each of its nodes. Then, we learn to predict affinities related to this graph, that ensures a decent salient object detection performance, when used with a spectral graph based foreground detection method. To accomplish this task, we modify convolutional kernel networks (CKNs) for graph affinity calculation, which were originally proposed to predict similarities between images. Subsequently, we employ a spectral graph based salient object detection method – Extended Quantum Cuts (EQCut) – using these graph affinities. We show that the salient object detection error of such a system is differentiable with respect to the parameters of the CKN. Therefore, the proposed system can be trained end-to-end by applying error backpropagation and CKN parameters can be learned for salient object detection task. The comparative evaluations over a large set of benchmark datasets indicate that the proposed method has an insignificant computational burden on, but significantly outperforms the baseline EQCut – which uses color affinities – and achieves a comparable performance level with the state-of-the-art in some performance measures.",
author = "Caglar Aytekin and Alexandros Iosifidis and Serkan Kiranyaz and Moncef Gabbouj",
note = "EXT={"}Kiranyaz, Serkan{"}",
year = "2017",
month = "4",
doi = "10.1016/j.patcog.2016.11.005",
language = "English",
volume = "64",
pages = "159--167",
journal = "Pattern Recognition",
issn = "0031-3203",
publisher = "ELSEVIER SCI LTD",

}

RIS (suitable for import to EndNote) - Lataa

TY - JOUR

T1 - Learning graph affinities for spectral graph-based salient object detection

AU - Aytekin, Caglar

AU - Iosifidis, Alexandros

AU - Kiranyaz, Serkan

AU - Gabbouj, Moncef

N1 - EXT="Kiranyaz, Serkan"

PY - 2017/4

Y1 - 2017/4

N2 - In this paper, we propose a novel method for learning graph affinities for salient object detection. First, we assume that a graph representation of an image is given with a predetermined connectivity rule and representative features for each of its nodes. Then, we learn to predict affinities related to this graph, that ensures a decent salient object detection performance, when used with a spectral graph based foreground detection method. To accomplish this task, we modify convolutional kernel networks (CKNs) for graph affinity calculation, which were originally proposed to predict similarities between images. Subsequently, we employ a spectral graph based salient object detection method – Extended Quantum Cuts (EQCut) – using these graph affinities. We show that the salient object detection error of such a system is differentiable with respect to the parameters of the CKN. Therefore, the proposed system can be trained end-to-end by applying error backpropagation and CKN parameters can be learned for salient object detection task. The comparative evaluations over a large set of benchmark datasets indicate that the proposed method has an insignificant computational burden on, but significantly outperforms the baseline EQCut – which uses color affinities – and achieves a comparable performance level with the state-of-the-art in some performance measures.

AB - In this paper, we propose a novel method for learning graph affinities for salient object detection. First, we assume that a graph representation of an image is given with a predetermined connectivity rule and representative features for each of its nodes. Then, we learn to predict affinities related to this graph, that ensures a decent salient object detection performance, when used with a spectral graph based foreground detection method. To accomplish this task, we modify convolutional kernel networks (CKNs) for graph affinity calculation, which were originally proposed to predict similarities between images. Subsequently, we employ a spectral graph based salient object detection method – Extended Quantum Cuts (EQCut) – using these graph affinities. We show that the salient object detection error of such a system is differentiable with respect to the parameters of the CKN. Therefore, the proposed system can be trained end-to-end by applying error backpropagation and CKN parameters can be learned for salient object detection task. The comparative evaluations over a large set of benchmark datasets indicate that the proposed method has an insignificant computational burden on, but significantly outperforms the baseline EQCut – which uses color affinities – and achieves a comparable performance level with the state-of-the-art in some performance measures.

U2 - 10.1016/j.patcog.2016.11.005

DO - 10.1016/j.patcog.2016.11.005

M3 - Article

VL - 64

SP - 159

EP - 167

JO - Pattern Recognition

JF - Pattern Recognition

SN - 0031-3203

ER -