Multi-Label Image Classification with Weak Correlation Prior
Image classification is vital and basic in many data analysis domains. Since real-world images generally contain multiple diverse semantic labels, it amounts to a typical multi-label classification problem. Traditional multi-label image classification relies on a large amount of training data with p...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Tsinghua University Press
2022-09-01
|
| Series: | CAAI Artificial Intelligence Research |
| Subjects: | |
| Online Access: | https://www.sciopen.com/article/10.26599/AIR.2022.9150006 |
| _version_ | 1797293759146229760 |
|---|---|
| author | Xiao Ouyang Ruidong Fan Hong Tao Chenping Hou |
| author_facet | Xiao Ouyang Ruidong Fan Hong Tao Chenping Hou |
| author_sort | Xiao Ouyang |
| collection | DOAJ |
| description | Image classification is vital and basic in many data analysis domains. Since real-world images generally contain multiple diverse semantic labels, it amounts to a typical multi-label classification problem. Traditional multi-label image classification relies on a large amount of training data with plenty of labels, which requires a lot of human and financial costs. By contrast, one can easily obtain a correlation matrix of concerned categories in current scene based on the historical image data in other application scenarios. How to perform image classification with only label correlation priors, without specific and costly annotated labels, is an important but rarely studied problem. In this paper, we propose a model to classify images with this kind of weak correlation prior. We use label correlation to recapitulate the sample similarity, employ the prior information to decompose the projection matrix when regressing the label indication matrix, and introduce the ℓ2,1 norm to select features for each image. Finally, experimental results on several image datasets demonstrate that the proposed model has distinct advantages over current state-of-the-art multi-label classification methods. |
| first_indexed | 2024-03-07T21:19:32Z |
| format | Article |
| id | doaj.art-02becd6170dc4f79963e309e47760076 |
| institution | Directory Open Access Journal |
| issn | 2097-194X |
| language | English |
| last_indexed | 2024-03-07T21:19:32Z |
| publishDate | 2022-09-01 |
| publisher | Tsinghua University Press |
| record_format | Article |
| series | CAAI Artificial Intelligence Research |
| spelling | doaj.art-02becd6170dc4f79963e309e477600762024-02-27T14:39:51ZengTsinghua University PressCAAI Artificial Intelligence Research2097-194X2022-09-0111799210.26599/AIR.2022.9150006Multi-Label Image Classification with Weak Correlation PriorXiao Ouyang0Ruidong Fan1Hong Tao2Chenping Hou3Department of Systems Science, National University of Defense Technology, Changsha 410073, ChinaDepartment of Systems Science, National University of Defense Technology, Changsha 410073, ChinaDepartment of Systems Science, National University of Defense Technology, Changsha 410073, ChinaDepartment of Systems Science, National University of Defense Technology, Changsha 410073, ChinaImage classification is vital and basic in many data analysis domains. Since real-world images generally contain multiple diverse semantic labels, it amounts to a typical multi-label classification problem. Traditional multi-label image classification relies on a large amount of training data with plenty of labels, which requires a lot of human and financial costs. By contrast, one can easily obtain a correlation matrix of concerned categories in current scene based on the historical image data in other application scenarios. How to perform image classification with only label correlation priors, without specific and costly annotated labels, is an important but rarely studied problem. In this paper, we propose a model to classify images with this kind of weak correlation prior. We use label correlation to recapitulate the sample similarity, employ the prior information to decompose the projection matrix when regressing the label indication matrix, and introduce the ℓ2,1 norm to select features for each image. Finally, experimental results on several image datasets demonstrate that the proposed model has distinct advantages over current state-of-the-art multi-label classification methods.https://www.sciopen.com/article/10.26599/AIR.2022.9150006image recognitionlabel correlationmulti-label classificationweakly-supervised learning |
| spellingShingle | Xiao Ouyang Ruidong Fan Hong Tao Chenping Hou Multi-Label Image Classification with Weak Correlation Prior CAAI Artificial Intelligence Research image recognition label correlation multi-label classification weakly-supervised learning |
| title | Multi-Label Image Classification with Weak Correlation Prior |
| title_full | Multi-Label Image Classification with Weak Correlation Prior |
| title_fullStr | Multi-Label Image Classification with Weak Correlation Prior |
| title_full_unstemmed | Multi-Label Image Classification with Weak Correlation Prior |
| title_short | Multi-Label Image Classification with Weak Correlation Prior |
| title_sort | multi label image classification with weak correlation prior |
| topic | image recognition label correlation multi-label classification weakly-supervised learning |
| url | https://www.sciopen.com/article/10.26599/AIR.2022.9150006 |
| work_keys_str_mv | AT xiaoouyang multilabelimageclassificationwithweakcorrelationprior AT ruidongfan multilabelimageclassificationwithweakcorrelationprior AT hongtao multilabelimageclassificationwithweakcorrelationprior AT chenpinghou multilabelimageclassificationwithweakcorrelationprior |