Details
| Originalsprache | Englisch |
|---|---|
| Titel des Sammelwerks | ICMR 2021 |
| Untertitel | Proceedings of the 2021 International Conference on Multimedia Retrieval |
| Seiten | 466-470 |
| Seitenumfang | 5 |
| ISBN (elektronisch) | 9781450384636 |
| Publikationsstatus | Veröffentlicht - 1 Sept. 2021 |
| Veranstaltung | 11th ACM International Conference on Multimedia Retrieval, ICMR 2021 - Taipei, Taiwan Dauer: 16 Nov. 2021 → 19 Nov. 2021 |
Publikationsreihe
| Name | ICMR 2021 - Proceedings of the 2021 International Conference on Multimedia Retrieval |
|---|
Abstract
Video summarization aims at generating a compact yet representative visual summary that conveys the essence of the original video. The advantage of unsupervised approaches is that they do not require human annotations to learn the summarization capability and generalize to a wider range of domains. Previous work relies on the same type of deep features, typically based on a model pre-trained on ImageNet data. Therefore, we propose to incorporate multiple feature sources with chunk and stride fusion to provide more information about the visual content. For a comprehensive evaluation on the two benchmarks TVSum and SumMe, we compare our method with four state-of-the-art approaches. Two of these approaches were implemented by ourselves to reproduce the reported results. Our evaluation shows that we obtain state-of-the-art results on both datasets while also highlighting the shortcomings of previous work with regard to the evaluation methodology. Finally, we perform error analysis on videos for the two benchmark datasets to summarize and spot the factors that lead to misclassifications.
ASJC Scopus Sachgebiete
- Informatik (insg.)
- Computergrafik und computergestütztes Design
- Informatik (insg.)
- Angewandte Informatik
- Informatik (insg.)
- Maschinelles Sehen und Mustererkennung
- Informatik (insg.)
- Mensch-Maschine-Interaktion
- Informatik (insg.)
- Software
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
ICMR 2021: Proceedings of the 2021 International Conference on Multimedia Retrieval. 2021. S. 466-470 (ICMR 2021 - Proceedings of the 2021 International Conference on Multimedia Retrieval).
Publikation: Beitrag in Buch/Bericht/Sammelwerk/Konferenzband › Aufsatz in Konferenzband › Forschung › Peer-Review
}
TY - GEN
T1 - Unsupervised Video Summarization via Multi-source Features
AU - Kanafani, Hussain
AU - Ghauri, Junaid Ahmed
AU - Hakimov, Sherzod
AU - Ewerth, Ralph
PY - 2021/9/1
Y1 - 2021/9/1
N2 - Video summarization aims at generating a compact yet representative visual summary that conveys the essence of the original video. The advantage of unsupervised approaches is that they do not require human annotations to learn the summarization capability and generalize to a wider range of domains. Previous work relies on the same type of deep features, typically based on a model pre-trained on ImageNet data. Therefore, we propose to incorporate multiple feature sources with chunk and stride fusion to provide more information about the visual content. For a comprehensive evaluation on the two benchmarks TVSum and SumMe, we compare our method with four state-of-the-art approaches. Two of these approaches were implemented by ourselves to reproduce the reported results. Our evaluation shows that we obtain state-of-the-art results on both datasets while also highlighting the shortcomings of previous work with regard to the evaluation methodology. Finally, we perform error analysis on videos for the two benchmark datasets to summarize and spot the factors that lead to misclassifications.
AB - Video summarization aims at generating a compact yet representative visual summary that conveys the essence of the original video. The advantage of unsupervised approaches is that they do not require human annotations to learn the summarization capability and generalize to a wider range of domains. Previous work relies on the same type of deep features, typically based on a model pre-trained on ImageNet data. Therefore, we propose to incorporate multiple feature sources with chunk and stride fusion to provide more information about the visual content. For a comprehensive evaluation on the two benchmarks TVSum and SumMe, we compare our method with four state-of-the-art approaches. Two of these approaches were implemented by ourselves to reproduce the reported results. Our evaluation shows that we obtain state-of-the-art results on both datasets while also highlighting the shortcomings of previous work with regard to the evaluation methodology. Finally, we perform error analysis on videos for the two benchmark datasets to summarize and spot the factors that lead to misclassifications.
KW - Deep learning
KW - Multi-source combination
KW - Multi-source fusion
KW - Unsupervised video summarization
KW - Video analysis
UR - http://www.scopus.com/inward/record.url?scp=85114886998&partnerID=8YFLogxK
U2 - https://doi.org/10.48550/arXiv.2105.12532
DO - https://doi.org/10.48550/arXiv.2105.12532
M3 - Conference contribution
AN - SCOPUS:85114886998
T3 - ICMR 2021 - Proceedings of the 2021 International Conference on Multimedia Retrieval
SP - 466
EP - 470
BT - ICMR 2021
T2 - 11th ACM International Conference on Multimedia Retrieval, ICMR 2021
Y2 - 16 November 2021 through 19 November 2021
ER -