Details
| Original language | English |
|---|---|
| Publication status | Published - 2006 |
| Externally published | Yes |
| Event | TREC Video Retrieval Evaluation, TRECVID 2006 - Gaithersburg, MD, United States Duration: 13 Nov 2006 → 14 Nov 2006 |
Conference
| Conference | TREC Video Retrieval Evaluation, TRECVID 2006 |
|---|---|
| Country/Territory | United States |
| City | Gaithersburg, MD |
| Period | 13 Nov 2006 → 14 Nov 2006 |
Abstract
In this paper, we summarize our results for the shot boundary detection task and the rushes task at TRECVID 2006. The shot boundary detection approach which was evaluated last year at TRECVID 2005 served as a basis for our experiments this year and was modified in several ways. First, we investigated different parameter settings for the unsupervised approach. Second, we experimented with the possibility to create an unsupervised ensemble that consists of several clusterings that have been obtained with different parameter settings. Our prototype for the rushes task consists of a summarization component and a retrieval component. Rushes videos are segmented on a sub-shot basis in order to separate redundant from non-redundant sequences within a shot. The summarization component is based on subshots clustering, and an appropriate visualization of clusters is presented to the user. The sub-shots are clustered with respect to a number of low-level and mid-level features, and they are visualized such that the user can navigate through these sub-shots. The retrieval component enables the user to search the rushes material automatically according to several features: camera motion, audio features (silence, speech, music, action, and background), speaker identity and interviews, shot sizes, face appearances, and by queries by example based on color and texture features.
ASJC Scopus subject areas
- Computer Science(all)
- Computer Graphics and Computer-Aided Design
- Computer Science(all)
- Computer Vision and Pattern Recognition
- Computer Science(all)
- Human-Computer Interaction
- Computer Science(all)
- Software
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
2006. Paper presented at TREC Video Retrieval Evaluation, TRECVID 2006, Gaithersburg, MD, United States.
Research output: Contribution to conference › Paper › Research › peer review
}
TY - CONF
T1 - University of Marburg at TRECVID 2006
T2 - TREC Video Retrieval Evaluation, TRECVID 2006
AU - Ewerth, Ralph
AU - Mühling, Markus
AU - Stadelmann, Thilo
AU - Qeli, Ermir
AU - Agel, Björn
AU - Seiler, Dominik
AU - Freisleben, Bernd
PY - 2006
Y1 - 2006
N2 - In this paper, we summarize our results for the shot boundary detection task and the rushes task at TRECVID 2006. The shot boundary detection approach which was evaluated last year at TRECVID 2005 served as a basis for our experiments this year and was modified in several ways. First, we investigated different parameter settings for the unsupervised approach. Second, we experimented with the possibility to create an unsupervised ensemble that consists of several clusterings that have been obtained with different parameter settings. Our prototype for the rushes task consists of a summarization component and a retrieval component. Rushes videos are segmented on a sub-shot basis in order to separate redundant from non-redundant sequences within a shot. The summarization component is based on subshots clustering, and an appropriate visualization of clusters is presented to the user. The sub-shots are clustered with respect to a number of low-level and mid-level features, and they are visualized such that the user can navigate through these sub-shots. The retrieval component enables the user to search the rushes material automatically according to several features: camera motion, audio features (silence, speech, music, action, and background), speaker identity and interviews, shot sizes, face appearances, and by queries by example based on color and texture features.
AB - In this paper, we summarize our results for the shot boundary detection task and the rushes task at TRECVID 2006. The shot boundary detection approach which was evaluated last year at TRECVID 2005 served as a basis for our experiments this year and was modified in several ways. First, we investigated different parameter settings for the unsupervised approach. Second, we experimented with the possibility to create an unsupervised ensemble that consists of several clusterings that have been obtained with different parameter settings. Our prototype for the rushes task consists of a summarization component and a retrieval component. Rushes videos are segmented on a sub-shot basis in order to separate redundant from non-redundant sequences within a shot. The summarization component is based on subshots clustering, and an appropriate visualization of clusters is presented to the user. The sub-shots are clustered with respect to a number of low-level and mid-level features, and they are visualized such that the user can navigate through these sub-shots. The retrieval component enables the user to search the rushes material automatically according to several features: camera motion, audio features (silence, speech, music, action, and background), speaker identity and interviews, shot sizes, face appearances, and by queries by example based on color and texture features.
UR - http://www.scopus.com/inward/record.url?scp=84905178040&partnerID=8YFLogxK
M3 - Paper
Y2 - 13 November 2006 through 14 November 2006
ER -