Tiny images
The human visual system is remarkably tolerant to degradations in image resolution: in a scene recognition task, human performance is similar whether $32 \times 32$ color images or multi-mega pixel images are used. With small images, even object recognition and segmentation is performed robustly by...
| Main Authors: | , , |
|---|---|
| Other Authors: | |
| Published: |
2007
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/1721.1/37291 |
| _version_ | 1826216743445463040 |
|---|---|
| author | Torralba, Antonio Fergus, Rob Freeman, William T. |
| author2 | William Freeman |
| author_facet | William Freeman Torralba, Antonio Fergus, Rob Freeman, William T. |
| author_sort | Torralba, Antonio |
| collection | MIT |
| description | The human visual system is remarkably tolerant to degradations in image resolution: in a scene recognition task, human performance is similar whether $32 \times 32$ color images or multi-mega pixel images are used. With small images, even object recognition and segmentation is performed robustly by the visual system, despite the object being unrecognizable in isolation. Motivated by these observations, we explore the space of 32x32 images using a database of 10^8 32x32 color images gathered from the Internet using image search engines. Each image is loosely labeled with one of the 70,399 non-abstract nouns in English, as listed in the Wordnet lexical database. Hence the image database represents a dense sampling of all object categories and scenes. With this dataset, we use nearest neighbor methods to perform objectrecognition across the 10^8 images. |
| first_indexed | 2024-09-23T16:52:25Z |
| id | mit-1721.1/37291 |
| institution | Massachusetts Institute of Technology |
| last_indexed | 2024-09-23T16:52:25Z |
| publishDate | 2007 |
| record_format | dspace |
| spelling | mit-1721.1/372912019-04-12T07:40:31Z Tiny images Torralba, Antonio Fergus, Rob Freeman, William T. William Freeman Vision Recognition Nearest neighbors methods Image databases The human visual system is remarkably tolerant to degradations in image resolution: in a scene recognition task, human performance is similar whether $32 \times 32$ color images or multi-mega pixel images are used. With small images, even object recognition and segmentation is performed robustly by the visual system, despite the object being unrecognizable in isolation. Motivated by these observations, we explore the space of 32x32 images using a database of 10^8 32x32 color images gathered from the Internet using image search engines. Each image is loosely labeled with one of the 70,399 non-abstract nouns in English, as listed in the Wordnet lexical database. Hence the image database represents a dense sampling of all object categories and scenes. With this dataset, we use nearest neighbor methods to perform objectrecognition across the 10^8 images. 2007-04-24T14:01:48Z 2007-04-24T14:01:48Z 2007-04-23 MIT-CSAIL-TR-2007-024 http://hdl.handle.net/1721.1/37291 Massachusetts Institute of Technology Computer Science and Artificial Intelligence Laboratory 9 p. application/postscript application/pdf |
| spellingShingle | Recognition Nearest neighbors methods Image databases Torralba, Antonio Fergus, Rob Freeman, William T. Tiny images |
| title | Tiny images |
| title_full | Tiny images |
| title_fullStr | Tiny images |
| title_full_unstemmed | Tiny images |
| title_short | Tiny images |
| title_sort | tiny images |
| topic | Recognition Nearest neighbors methods Image databases |
| url | http://hdl.handle.net/1721.1/37291 |
| work_keys_str_mv | AT torralbaantonio tinyimages AT fergusrob tinyimages AT freemanwilliamt tinyimages |