Tampere University of Technology

TUTCRIS Research Portal

Learning to Classify Fine-Grained Categories with Privileged Visual-Semantic Misalignment

Research output: Contribution to journalArticleScientificpeer-review

Standard

Learning to Classify Fine-Grained Categories with Privileged Visual-Semantic Misalignment. / Chen, Ke; Zhang, Zhaoxiang.

In: IEEE Transactions on Big Data, Vol. 3, No. 1, 2016, p. 37-43.

Research output: Contribution to journalArticleScientificpeer-review

Harvard

APA

Vancouver

Author

Chen, Ke ; Zhang, Zhaoxiang. / Learning to Classify Fine-Grained Categories with Privileged Visual-Semantic Misalignment. In: IEEE Transactions on Big Data. 2016 ; Vol. 3, No. 1. pp. 37-43.

Bibtex - Download

@article{d732702358774f92b1a4758ebdce065e,
title = "Learning to Classify Fine-Grained Categories with Privileged Visual-Semantic Misalignment",
abstract = "Image categorisation is an active yet challenging research topic in computer vision, which is to classify the images according to their semantic content. Recently, fine-grained object categorisation has attracted wide attention and remains difficult due to feature inconsistency caused by smaller inter-class and larger intra-class variation as well as large varying poses. Most of the existing frameworks focused on exploiting a more discriminative imagery representation or developing a more robust classification framework to mitigate the suffering. The concern has recently been paid to discovering the dependency across fine-grained class labels based on Convolutional Neural Networks. Encouraged by the success of semantic label embedding to discover the fine-grained class labels’ correlation, this paper exploits the misalignment between visual feature space and semantic label embedding space and incorporates it as a privileged information into a cost-sensitive learning framework. Owing to capturing both the variation of imagery feature representation and also the label correlation in the semantic label embedding space, such a visual-semantic misalignment can be employed to reflect the importance of instances, which is more informative that conventional cost-sensitivities. Experiment results demonstrate the effectiveness of the proposed framework on public fine-grained benchmarks with achieving superior performance to state-of-the-arts.",
author = "Ke Chen and Zhaoxiang Zhang",
year = "2016",
doi = "10.1109/TBDATA.2016.2602231",
language = "English",
volume = "3",
pages = "37--43",
journal = "IEEE Transactions on Big Data",
issn = "2332-7790",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "1",

}

RIS (suitable for import to EndNote) - Download

TY - JOUR

T1 - Learning to Classify Fine-Grained Categories with Privileged Visual-Semantic Misalignment

AU - Chen, Ke

AU - Zhang, Zhaoxiang

PY - 2016

Y1 - 2016

N2 - Image categorisation is an active yet challenging research topic in computer vision, which is to classify the images according to their semantic content. Recently, fine-grained object categorisation has attracted wide attention and remains difficult due to feature inconsistency caused by smaller inter-class and larger intra-class variation as well as large varying poses. Most of the existing frameworks focused on exploiting a more discriminative imagery representation or developing a more robust classification framework to mitigate the suffering. The concern has recently been paid to discovering the dependency across fine-grained class labels based on Convolutional Neural Networks. Encouraged by the success of semantic label embedding to discover the fine-grained class labels’ correlation, this paper exploits the misalignment between visual feature space and semantic label embedding space and incorporates it as a privileged information into a cost-sensitive learning framework. Owing to capturing both the variation of imagery feature representation and also the label correlation in the semantic label embedding space, such a visual-semantic misalignment can be employed to reflect the importance of instances, which is more informative that conventional cost-sensitivities. Experiment results demonstrate the effectiveness of the proposed framework on public fine-grained benchmarks with achieving superior performance to state-of-the-arts.

AB - Image categorisation is an active yet challenging research topic in computer vision, which is to classify the images according to their semantic content. Recently, fine-grained object categorisation has attracted wide attention and remains difficult due to feature inconsistency caused by smaller inter-class and larger intra-class variation as well as large varying poses. Most of the existing frameworks focused on exploiting a more discriminative imagery representation or developing a more robust classification framework to mitigate the suffering. The concern has recently been paid to discovering the dependency across fine-grained class labels based on Convolutional Neural Networks. Encouraged by the success of semantic label embedding to discover the fine-grained class labels’ correlation, this paper exploits the misalignment between visual feature space and semantic label embedding space and incorporates it as a privileged information into a cost-sensitive learning framework. Owing to capturing both the variation of imagery feature representation and also the label correlation in the semantic label embedding space, such a visual-semantic misalignment can be employed to reflect the importance of instances, which is more informative that conventional cost-sensitivities. Experiment results demonstrate the effectiveness of the proposed framework on public fine-grained benchmarks with achieving superior performance to state-of-the-arts.

U2 - 10.1109/TBDATA.2016.2602231

DO - 10.1109/TBDATA.2016.2602231

M3 - Article

VL - 3

SP - 37

EP - 43

JO - IEEE Transactions on Big Data

JF - IEEE Transactions on Big Data

SN - 2332-7790

IS - 1

ER -