Learning object boundary detection from motion data
This paper describes the initial results of a project to create a self-supervised algorithm for learning object segmentation from video data. Developmental psychology and computational experience have demonstrated that the motion segmentation of objects is a simpler, more primitive process than the...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | en_US |
| Published: |
2003
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/1721.1/3870 |
| _version_ | 1826194823733837824 |
|---|---|
| author | Ross, Michael G. Kaelbling, Leslie P. |
| author_facet | Ross, Michael G. Kaelbling, Leslie P. |
| author_sort | Ross, Michael G. |
| collection | MIT |
| description | This paper describes the initial results of a project to create a self-supervised algorithm for learning object segmentation from video data. Developmental psychology and computational experience have demonstrated that the motion segmentation of objects is a simpler, more primitive process than the detection of object boundaries by static image cues. Therefore, motion information provides a plausible supervision signal for learning the static boundary detection task and for evaluating performance on a test set. A video camera and previously developed background subtraction algorithms can automatically produce a large database of motion-segmented images for minimal cost. The purpose of this work is to use the information in such a database to learn how to detect the object boundaries in novel images using static information, such as color, texture, and shape. |
| first_indexed | 2024-09-23T10:02:21Z |
| format | Article |
| id | mit-1721.1/3870 |
| institution | Massachusetts Institute of Technology |
| language | en_US |
| last_indexed | 2024-09-23T10:02:21Z |
| publishDate | 2003 |
| record_format | dspace |
| spelling | mit-1721.1/38702019-04-12T11:15:06Z Learning object boundary detection from motion data Ross, Michael G. Kaelbling, Leslie P. machine learning self-supervised algorithm motion segmentation object boundary detection This paper describes the initial results of a project to create a self-supervised algorithm for learning object segmentation from video data. Developmental psychology and computational experience have demonstrated that the motion segmentation of objects is a simpler, more primitive process than the detection of object boundaries by static image cues. Therefore, motion information provides a plausible supervision signal for learning the static boundary detection task and for evaluating performance on a test set. A video camera and previously developed background subtraction algorithms can automatically produce a large database of motion-segmented images for minimal cost. The purpose of this work is to use the information in such a database to learn how to detect the object boundaries in novel images using static information, such as color, texture, and shape. Singapore-MIT Alliance (SMA) 2003-12-13T20:13:43Z 2003-12-13T20:13:43Z 2004-01 Article http://hdl.handle.net/1721.1/3870 en_US Computer Science (CS); 1234090 bytes application/pdf application/pdf |
| spellingShingle | machine learning self-supervised algorithm motion segmentation object boundary detection Ross, Michael G. Kaelbling, Leslie P. Learning object boundary detection from motion data |
| title | Learning object boundary detection from motion data |
| title_full | Learning object boundary detection from motion data |
| title_fullStr | Learning object boundary detection from motion data |
| title_full_unstemmed | Learning object boundary detection from motion data |
| title_short | Learning object boundary detection from motion data |
| title_sort | learning object boundary detection from motion data |
| topic | machine learning self-supervised algorithm motion segmentation object boundary detection |
| url | http://hdl.handle.net/1721.1/3870 |
| work_keys_str_mv | AT rossmichaelg learningobjectboundarydetectionfrommotiondata AT kaelblinglesliep learningobjectboundarydetectionfrommotiondata |