Proceedings of the 15th International Workshop on Content-Based Multimedia Indexing 2017
DOI: 10.1145/3095713.3095729
|View full text |Cite
|
Sign up to set email alerts
|

Tag Propagation Approaches within Speaking Face Graphs for Multimodal Person Discovery

Abstract: The indexing of broadcast TV archives is a current problem in multimedia research. As the size of these databases grows continuously, meaningful features are needed to describe and connect their elements efficiently, such as the identification of speaking faces. In this context, this paper focuses on two approaches for unsupervised person discovery. Initial tagging of speaking faces is provided by an OCR-based method, and these tags propagate through a graph model based on audiovisual relations between speakin… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1

Citation Types

0
4
0

Year Published

2020
2020
2020
2020

Publication Types

Select...
1

Relationship

1
0

Authors

Journals

citations
Cited by 1 publication
(4 citation statements)
references
References 10 publications
0
4
0
Order By: Relevance
“…In a previous work [14], we have developed a label propagation based on minimum spanning trees; and, according to [13], a QFZ hierarchy and a minimum spanning tree (MST) are equivalent, thus a QFZ hierarchy can be computed from a component tree or one can compute an MST to produce the same hierarchy since both are equivalent. Here, we take advantage of this equivalence to create an efficient implementation of the QFZ label propagation, using the procedure for creating component trees in quasi-linear time [36].…”
Section: Hierarchical Multi-label Propagationmentioning
confidence: 99%
See 3 more Smart Citations
“…In a previous work [14], we have developed a label propagation based on minimum spanning trees; and, according to [13], a QFZ hierarchy and a minimum spanning tree (MST) are equivalent, thus a QFZ hierarchy can be computed from a component tree or one can compute an MST to produce the same hierarchy since both are equivalent. Here, we take advantage of this equivalence to create an efficient implementation of the QFZ label propagation, using the procedure for creating component trees in quasi-linear time [36].…”
Section: Hierarchical Multi-label Propagationmentioning
confidence: 99%
“…In this work, we extend the study of a semi-supervised hierarchical label propagation method to propagate initially extracted names to all occurrences of the same person in a video, as proposed in our previous work [14]. This label propagation approach is inspired by a seeded hierarchical image segmentation method [1,42,43].…”
Section: Introductionmentioning
confidence: 99%
See 2 more Smart Citations