A Multi-Label Image Classification Method based on Label Correlation Learning Network

[Purposes] To meet the challenges posed by label feature confusions and limitations in label relationships in multi-label image classification tasks, a novel approach to multi-label image classification based on label correlation learning network (MLLCLN) is presented in this work. [Methods] MLLCLN...

Full description

Saved in:
Bibliographic Details
Main Authors: WANG Lufang, ZHANG Haiyun
Format: Article
Language:English
Published: Editorial Office of Journal of Taiyuan University of Technology 2024-11-01
Series:Taiyuan Ligong Daxue xuebao
Subjects:
Online Access:https://tyutjournal.tyut.edu.cn/englishpaper/show-2358.html
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846150334566629376
author WANG Lufang
ZHANG Haiyun
author_facet WANG Lufang
ZHANG Haiyun
author_sort WANG Lufang
collection DOAJ
description [Purposes] To meet the challenges posed by label feature confusions and limitations in label relationships in multi-label image classification tasks, a novel approach to multi-label image classification based on label correlation learning network (MLLCLN) is presented in this work. [Methods] MLLCLN adopts the methods of masked attention approach and multi-head self-attention mechanism. In the masked attention approach, the label features generated by masking the attention mechanism with state word vectors corresponding to the real labels in the image, allowing the model to obtain more contextual information and mitigating the issue of attention overlap in the attention regions. This strategy effectively alleviate the issue of label feature confusion. Moreover, a label correlation learning network is devised, which comprises multiple layers of multi-head attention mechanisms and a graph neural network. On the other hand, the multi-head self-attention mechanism enables the learning of local label relationships according to the label features, while the graph neural network incorporates the widely adopted ML-GCN method to guide the model in considering global label relationships simultaneously, mitigating the issue of false predictions in models caused by the limitations of label relationships. [Findings] The experimental results of MLLCLN on the public datasets MSCOCO2014 and VOC2007 demonstrate its superior performance, achieving classification accuracies of 84.4% and 96.0%, respectively. This provides a novel approach to multi-label image classification.
format Article
id doaj-art-9fb6e007df8141ebb25833003ca5a87b
institution Kabale University
issn 1007-9432
language English
publishDate 2024-11-01
publisher Editorial Office of Journal of Taiyuan University of Technology
record_format Article
series Taiyuan Ligong Daxue xuebao
spelling doaj-art-9fb6e007df8141ebb25833003ca5a87b2024-11-29T03:39:51ZengEditorial Office of Journal of Taiyuan University of TechnologyTaiyuan Ligong Daxue xuebao1007-94322024-11-015561097110610.16355/j.tyut.1007-9432.202404461007-9432(2024)06-1097-10A Multi-Label Image Classification Method based on Label Correlation Learning NetworkWANG Lufang0ZHANG Haiyun1Experimental Training Center, Shanxi University of Finance and Economics, Taiyuan 030031, Chinaa. Institute of Big Data Science and Industry, 2b. School of Computer and Information Technology, Shanxi University, Taiyuan 030006, China[Purposes] To meet the challenges posed by label feature confusions and limitations in label relationships in multi-label image classification tasks, a novel approach to multi-label image classification based on label correlation learning network (MLLCLN) is presented in this work. [Methods] MLLCLN adopts the methods of masked attention approach and multi-head self-attention mechanism. In the masked attention approach, the label features generated by masking the attention mechanism with state word vectors corresponding to the real labels in the image, allowing the model to obtain more contextual information and mitigating the issue of attention overlap in the attention regions. This strategy effectively alleviate the issue of label feature confusion. Moreover, a label correlation learning network is devised, which comprises multiple layers of multi-head attention mechanisms and a graph neural network. On the other hand, the multi-head self-attention mechanism enables the learning of local label relationships according to the label features, while the graph neural network incorporates the widely adopted ML-GCN method to guide the model in considering global label relationships simultaneously, mitigating the issue of false predictions in models caused by the limitations of label relationships. [Findings] The experimental results of MLLCLN on the public datasets MSCOCO2014 and VOC2007 demonstrate its superior performance, achieving classification accuracies of 84.4% and 96.0%, respectively. This provides a novel approach to multi-label image classification.https://tyutjournal.tyut.edu.cn/englishpaper/show-2358.htmlmulti-head self-attentionmulti-label image classificationattention mechanismadaptive weightconvolutional neural networks
spellingShingle WANG Lufang
ZHANG Haiyun
A Multi-Label Image Classification Method based on Label Correlation Learning Network
Taiyuan Ligong Daxue xuebao
multi-head self-attention
multi-label image classification
attention mechanism
adaptive weight
convolutional neural networks
title A Multi-Label Image Classification Method based on Label Correlation Learning Network
title_full A Multi-Label Image Classification Method based on Label Correlation Learning Network
title_fullStr A Multi-Label Image Classification Method based on Label Correlation Learning Network
title_full_unstemmed A Multi-Label Image Classification Method based on Label Correlation Learning Network
title_short A Multi-Label Image Classification Method based on Label Correlation Learning Network
title_sort multi label image classification method based on label correlation learning network
topic multi-head self-attention
multi-label image classification
attention mechanism
adaptive weight
convolutional neural networks
url https://tyutjournal.tyut.edu.cn/englishpaper/show-2358.html
work_keys_str_mv AT wanglufang amultilabelimageclassificationmethodbasedonlabelcorrelationlearningnetwork
AT zhanghaiyun amultilabelimageclassificationmethodbasedonlabelcorrelationlearningnetwork
AT wanglufang multilabelimageclassificationmethodbasedonlabelcorrelationlearningnetwork
AT zhanghaiyun multilabelimageclassificationmethodbasedonlabelcorrelationlearningnetwork