Segmenting Scenes by Matching Image Composites
In this paper, we investigate how, given an image, similar images sharing the same global description can help with unsupervised scene segmentation. In contrast to recent work in semantic alignment of scenes, we allow an input image to be explainedby partial matches of similar scenes. This allows fo...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
2021
|
| Online Access: | https://hdl.handle.net/1721.1/137534 |
| _version_ | 1826203718345818112 |
|---|---|
| author | Russell, Bryan C. Efros, Alexei A. Sivic, Josef Freeman, William T. Zisserman, Andrew |
| author_facet | Russell, Bryan C. Efros, Alexei A. Sivic, Josef Freeman, William T. Zisserman, Andrew |
| author_sort | Russell, Bryan C. |
| collection | MIT |
| description | In this paper, we investigate how, given an image, similar images sharing the same global description can help with unsupervised scene segmentation. In contrast to recent work in semantic alignment of scenes, we allow an input image to be explainedby partial matches of similar scenes. This allows for a better explanation of the input scenes. We perform MRF-based segmentation that optimizes over matches, while respecting boundary information. The recovered segments are then used to re-query a large database of images to retrieve better matches for the target regions. We show improved performance in detecting the principal occluding and contact boundaries for the scene over previous methods on data gathered from the LabelMe database. |
| first_indexed | 2024-09-23T12:42:17Z |
| format | Article |
| id | mit-1721.1/137534 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T12:42:17Z |
| publishDate | 2021 |
| record_format | dspace |
| spelling | mit-1721.1/1375342021-11-06T03:22:34Z Segmenting Scenes by Matching Image Composites Russell, Bryan C. Efros, Alexei A. Sivic, Josef Freeman, William T. Zisserman, Andrew In this paper, we investigate how, given an image, similar images sharing the same global description can help with unsupervised scene segmentation. In contrast to recent work in semantic alignment of scenes, we allow an input image to be explainedby partial matches of similar scenes. This allows for a better explanation of the input scenes. We perform MRF-based segmentation that optimizes over matches, while respecting boundary information. The recovered segments are then used to re-query a large database of images to retrieve better matches for the target regions. We show improved performance in detecting the principal occluding and contact boundaries for the scene over previous methods on data gathered from the LabelMe database. 2021-11-05T16:17:02Z 2021-11-05T16:17:02Z 2009 2019-05-28T13:01:48Z Article http://purl.org/eprint/type/ConferencePaper https://hdl.handle.net/1721.1/137534 Russell, Bryan C., Efros, Alexei A., Sivic, Josef, Freeman, William T. and Zisserman, Andrew. 2009. "Segmenting Scenes by Matching Image Composites." en https://papers.nips.cc/paper/3718-segmenting-scenes-by-matching-image-composites Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use. application/pdf Neural Information Processing Systems (NIPS) |
| spellingShingle | Russell, Bryan C. Efros, Alexei A. Sivic, Josef Freeman, William T. Zisserman, Andrew Segmenting Scenes by Matching Image Composites |
| title | Segmenting Scenes by Matching Image Composites |
| title_full | Segmenting Scenes by Matching Image Composites |
| title_fullStr | Segmenting Scenes by Matching Image Composites |
| title_full_unstemmed | Segmenting Scenes by Matching Image Composites |
| title_short | Segmenting Scenes by Matching Image Composites |
| title_sort | segmenting scenes by matching image composites |
| url | https://hdl.handle.net/1721.1/137534 |
| work_keys_str_mv | AT russellbryanc segmentingscenesbymatchingimagecomposites AT efrosalexeia segmentingscenesbymatchingimagecomposites AT sivicjosef segmentingscenesbymatchingimagecomposites AT freemanwilliamt segmentingscenesbymatchingimagecomposites AT zissermanandrew segmentingscenesbymatchingimagecomposites |