Multi video summarization using query based deep optimization algorithm

The popularity of online video-sharing platforms has fuelled demand for systems that can quickly browse, extract, and summarise video information. Nowadays, numerous automatic multi-video summarization (MVS) techniques have come into existence. The existing MVS approach, on the other hand, produces...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:International journal of machine learning and cybernetics Jg. 14; H. 10; S. 3591 - 3606
Hauptverfasser: Ansari, Shaharyar Alam, Zafar, Aasim
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Berlin/Heidelberg Springer Berlin Heidelberg 01.10.2023
Springer Nature B.V
Schlagworte:
Algorithms
Artificial Intelligence
Cameras
Complex Systems
Computational Intelligence
Control
Engineering
Frames (data processing)
Mechatronics
Object recognition
Optimization
Optimization techniques
Original Article
Pattern Recognition
Queries
Robotics
Surveillance
Systems Biology
Video data
Key frame selection
Query based summarization
Multi-video summarization
African vulture optimization algorithm
Object detection
ISSN:1868-8071, 1868-808X
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The popularity of online video-sharing platforms has fuelled demand for systems that can quickly browse, extract, and summarise video information. Nowadays, numerous automatic multi-video summarization (MVS) techniques have come into existence. The existing MVS approach, on the other hand, produces summarised video with a lot of unimportant and duplicate frames. It also arranges frames in a meaningless manner. To solve these problems in MVS, Query-based Deep African Vulture Learning (QDAVOL) is proposed in this paper. It uses tag information and web images searched by the query as important information to identify the query intent. An event-based object detection and grouping (EODG) technique is used to assign keyframes to groups of specific events relevant with the query. In addition, we introduce the African vulture optimization algorithm (AVOA) for the efficient key frame selection. Moreover, we have also developed a similarity-based frame closeness (SFC) technique to provide more comprehensible summary. Experimental results demonstrate that the proposed framework outperforms existing approaches in terms of precision (0.765), recall (0.845), and average F-score (0.774) on MVS1K dataset.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1868-8071
1868-808X
DOI:10.1007/s13042-023-01852-3