Scene Segmentation with Low-Dimensional Semantic Representations and Conditional Random Fields
<p/> <p>This paper presents a fast, precise, and highly scalable semantic segmentation algorithm that incorporates several kinds of local appearance features, example-based spatial layout priors, and neighborhood-level and global contextual information. The method works at the level of i...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SpringerOpen
2010-01-01
|
| Series: | EURASIP Journal on Advances in Signal Processing |
| Online Access: | http://asp.eurasipjournals.com/content/2010/196036 |
| _version_ | 1819330655530516480 |
|---|---|
| author | Triggs Bill Xia Gui-Song Yang Wen Dai Dengxin |
| author_facet | Triggs Bill Xia Gui-Song Yang Wen Dai Dengxin |
| author_sort | Triggs Bill |
| collection | DOAJ |
| description | <p/> <p>This paper presents a fast, precise, and highly scalable semantic segmentation algorithm that incorporates several kinds of local appearance features, example-based spatial layout priors, and neighborhood-level and global contextual information. The method works at the level of image patches. In the first stage, codebook-based local appearance features are regularized and reduced in dimension using latent topic models, combined with spatial pyramid matching based spatial layout features, and fed into logistic regression classifiers to produce an initial patch level labeling. In the second stage, these labels are combined with patch-neighborhood and global aggregate features using either a second layer of Logistic Regression or a Conditional Random Field. Finally, the patch-level results are refined to pixel level using MRF or over-segmentation based methods. The CRF is trained using a fast Maximum Margin approach. Comparative experiments on four multi-class segmentation datasets show that each of the above elements improves the results, leading to a scalable algorithm that is both faster and more accurate than existing patch-level approaches.</p> |
| first_indexed | 2024-12-24T14:01:59Z |
| format | Article |
| id | doaj.art-5b1ad6ab461e472d9445ca47ed57db1e |
| institution | Directory Open Access Journal |
| issn | 1687-6172 1687-6180 |
| language | English |
| last_indexed | 2024-12-24T14:01:59Z |
| publishDate | 2010-01-01 |
| publisher | SpringerOpen |
| record_format | Article |
| series | EURASIP Journal on Advances in Signal Processing |
| spelling | doaj.art-5b1ad6ab461e472d9445ca47ed57db1e2022-12-21T16:52:27ZengSpringerOpenEURASIP Journal on Advances in Signal Processing1687-61721687-61802010-01-0120101196036Scene Segmentation with Low-Dimensional Semantic Representations and Conditional Random FieldsTriggs BillXia Gui-SongYang WenDai Dengxin<p/> <p>This paper presents a fast, precise, and highly scalable semantic segmentation algorithm that incorporates several kinds of local appearance features, example-based spatial layout priors, and neighborhood-level and global contextual information. The method works at the level of image patches. In the first stage, codebook-based local appearance features are regularized and reduced in dimension using latent topic models, combined with spatial pyramid matching based spatial layout features, and fed into logistic regression classifiers to produce an initial patch level labeling. In the second stage, these labels are combined with patch-neighborhood and global aggregate features using either a second layer of Logistic Regression or a Conditional Random Field. Finally, the patch-level results are refined to pixel level using MRF or over-segmentation based methods. The CRF is trained using a fast Maximum Margin approach. Comparative experiments on four multi-class segmentation datasets show that each of the above elements improves the results, leading to a scalable algorithm that is both faster and more accurate than existing patch-level approaches.</p>http://asp.eurasipjournals.com/content/2010/196036 |
| spellingShingle | Triggs Bill Xia Gui-Song Yang Wen Dai Dengxin Scene Segmentation with Low-Dimensional Semantic Representations and Conditional Random Fields EURASIP Journal on Advances in Signal Processing |
| title | Scene Segmentation with Low-Dimensional Semantic Representations and Conditional Random Fields |
| title_full | Scene Segmentation with Low-Dimensional Semantic Representations and Conditional Random Fields |
| title_fullStr | Scene Segmentation with Low-Dimensional Semantic Representations and Conditional Random Fields |
| title_full_unstemmed | Scene Segmentation with Low-Dimensional Semantic Representations and Conditional Random Fields |
| title_short | Scene Segmentation with Low-Dimensional Semantic Representations and Conditional Random Fields |
| title_sort | scene segmentation with low dimensional semantic representations and conditional random fields |
| url | http://asp.eurasipjournals.com/content/2010/196036 |
| work_keys_str_mv | AT triggsbill scenesegmentationwithlowdimensionalsemanticrepresentationsandconditionalrandomfields AT xiaguisong scenesegmentationwithlowdimensionalsemanticrepresentationsandconditionalrandomfields AT yangwen scenesegmentationwithlowdimensionalsemanticrepresentationsandconditionalrandomfields AT daidengxin scenesegmentationwithlowdimensionalsemanticrepresentationsandconditionalrandomfields |