Learning image similarity from Flickr groups using fast kernel machines
Measuring image similarity is a central topic in computer vision. In this paper, we propose to measure image similarity by learning from the online Flickr image groups. We do so by: Choosing 103 Flickr groups, building a one-versus-all multiclass classifier to classify test images into a group, taki...
| Главные авторы: | , , |
|---|---|
| Другие авторы: | |
| Формат: | Journal Article |
| Язык: | English |
| Опубликовано: |
2013
|
| Предметы: | |
| Online-ссылка: | https://hdl.handle.net/10356/99138 http://hdl.handle.net/10220/13494 |
| _version_ | 1826114562182610944 |
|---|---|
| author | Wang, Gang Hoiem, Derek Forsyth, David |
| author2 | School of Electrical and Electronic Engineering |
| author_facet | School of Electrical and Electronic Engineering Wang, Gang Hoiem, Derek Forsyth, David |
| author_sort | Wang, Gang |
| collection | NTU |
| description | Measuring image similarity is a central topic in computer vision. In this paper, we propose to measure image similarity by learning from the online Flickr image groups. We do so by: Choosing 103 Flickr groups, building a one-versus-all multiclass classifier to classify test images into a group, taking the set of responses of the classifiers as features, calculating the distance between feature vectors to measure image similarity. Experimental results on the Corel dataset and the PASCAL VOC 2007 dataset show that our approach performs better on image matching, retrieval, and classification than using conventional visual features. To build our similarity measure, we need one-versus-all classifiers that are accurate and can be trained quickly on very large quantities of data. We adopt an SVM classifier with a histogram intersection kernel. We describe a novel fast training algorithm for this classifier: the Stochastic Intersection Kernel MAchine (SIKMA) training algorithm. This method can produce a kernel classifier that is more accurate than a linear classifier on tens of thousands of examples in minutes. |
| first_indexed | 2024-10-01T03:41:08Z |
| format | Journal Article |
| id | ntu-10356/99138 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T03:41:08Z |
| publishDate | 2013 |
| record_format | dspace |
| spelling | ntu-10356/991382020-03-07T14:00:32Z Learning image similarity from Flickr groups using fast kernel machines Wang, Gang Hoiem, Derek Forsyth, David School of Electrical and Electronic Engineering DRNTU::Engineering::Electrical and electronic engineering Measuring image similarity is a central topic in computer vision. In this paper, we propose to measure image similarity by learning from the online Flickr image groups. We do so by: Choosing 103 Flickr groups, building a one-versus-all multiclass classifier to classify test images into a group, taking the set of responses of the classifiers as features, calculating the distance between feature vectors to measure image similarity. Experimental results on the Corel dataset and the PASCAL VOC 2007 dataset show that our approach performs better on image matching, retrieval, and classification than using conventional visual features. To build our similarity measure, we need one-versus-all classifiers that are accurate and can be trained quickly on very large quantities of data. We adopt an SVM classifier with a histogram intersection kernel. We describe a novel fast training algorithm for this classifier: the Stochastic Intersection Kernel MAchine (SIKMA) training algorithm. This method can produce a kernel classifier that is more accurate than a linear classifier on tens of thousands of examples in minutes. 2013-09-16T07:56:13Z 2019-12-06T20:03:46Z 2013-09-16T07:56:13Z 2019-12-06T20:03:46Z 2012 2012 Journal Article Wang, G., Hoiem, D., & Forsyth, D. (2012). Learning Image Similarity from Flickr Groups Using Fast Kernel Machines. IEEE Transactions on Pattern Analysis and Machine Intelligence, 34(11), 2177-2188. 0162-8828 https://hdl.handle.net/10356/99138 http://hdl.handle.net/10220/13494 10.1109/TPAMI.2012.29 en IEEE transactions on pattern analysis and machine intelligence © 2012 IEEE |
| spellingShingle | DRNTU::Engineering::Electrical and electronic engineering Wang, Gang Hoiem, Derek Forsyth, David Learning image similarity from Flickr groups using fast kernel machines |
| title | Learning image similarity from Flickr groups using fast kernel machines |
| title_full | Learning image similarity from Flickr groups using fast kernel machines |
| title_fullStr | Learning image similarity from Flickr groups using fast kernel machines |
| title_full_unstemmed | Learning image similarity from Flickr groups using fast kernel machines |
| title_short | Learning image similarity from Flickr groups using fast kernel machines |
| title_sort | learning image similarity from flickr groups using fast kernel machines |
| topic | DRNTU::Engineering::Electrical and electronic engineering |
| url | https://hdl.handle.net/10356/99138 http://hdl.handle.net/10220/13494 |
| work_keys_str_mv | AT wanggang learningimagesimilarityfromflickrgroupsusingfastkernelmachines AT hoiemderek learningimagesimilarityfromflickrgroupsusingfastkernelmachines AT forsythdavid learningimagesimilarityfromflickrgroupsusingfastkernelmachines |