# MPEG7 Video Retrieval using Bayesian Networks Luis M

MPEG-7 Video Retrieval using Bayesian Networks Luis M. de Campos Juan M. Fernández-Luna Juan F. Guadix Departamento de Ciencias de la Computación e Inteligencia Artificial E. T. S. I. Informática Universidad de Granada MPEG-7 video retrieval using Bayesian Networks

Introduction Brief overview about our work on the design of a 4 search engine based on Bayesian Networks 4 to retrieve MPEG-7 videos 4 using their text annotations. 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 2

Overview § Preliminaries: introduction to • Information Retrieval • MPEG-7 standard • Bayesian Networks § MPEG-7 video retrieval models based on Bayesian networks. 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 3

Preliminaries (I) – Information Retrieval is concerned with the representation, storage, organisation and accessing of information items. Indexing + Querying + Retrieval 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 4

Preliminaries (II) – MPEG-7 Multimedia Content Description Interface Standard to describe multimedia content using metadata: • The content of a multimedia file: concepts, objects in movement, who is speaking, . . . • Aspects related to the management of the content, i. e. , duration, structure, format and size of the file, number of frames per shot, . . . Tools: • Descriptors, Schemes, Data Definition Language. 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 5

Preliminaries (III) – MPEG-7 • Descriptors: elements, data representation. (Time to represent a duration, histogram to represent a colour or a string to represent a title) • Schemes: structure and semantic of the relationships among elements. (A film divided into scenes and shots, including textual description in the scene level and description about colour, movement and audio amplitude in the shot level) • DDL (Data Definition Language): Language to extend or modify the previous set of tools. It is a variety of XML Schema. Therefore, descriptions files are XML files. 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 6

Preliminaries (IV) – MPEG-7 <? xml version="1. 0" encoding="UTF-8"? > <Video. Segment id="shot 1_2"> <Mpeg 7> <Media. Time> <Description xsi: type="Content. Entity. Type"> <Media. Time. Point>T 00: 03: 22112 F 30000</Media. Time. Point> <Multimedia. Content xsi: type="Video. Type"> <Media. Duration>PT 9 S 18288 N 30000 F</Media. Duration> <Video id="1"> </Media. Time> <Text. Annotation confidence="0. 500000"> <Media. Time. Point>T 00: 00: 0 F 30000</Media. Time. Point> <Free. Text. Annotation> <Media. Duration>PT 16 M 33 S 11772 N 30000 F</Media. Duration> Collin Powell is speaking about the USA position </Media. Time> in the Iraq crisis. <Temporal. Decomposition gap="false" overlap="false"> <Video. Segment id="shot 1_1"> </Free. Text. Annotation> </Text. Annotation> <Media. Time> </Video. Segment> <Media. Time. Point>T 00: 00: 0 F 30000</Media. Time. Point> <Media. Duration>PT 3 S 22112 N 30000 F</Media. Duration> </Temporal. Decomposition> </Video> </Media. Time> </Multimedia. Content> <Text. Annotation confidence="0. 500000"> </Description> <Free. Text. Annotation> A tv presenter is reporting information about a meeting of the security council in UN. </Free. Text. Annotation> </Mpeg 7> </Text. Annotation> </Video. Segment> 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 7

Preliminaries (V) – MPEG-7 From the point of view of IR, the structure of a video is seen conceptually: Vídeo Scene 1 Shot 2 Scene 2 Shot 3 Shot 4 Scene 3 Shot 5 Shot 6 Frame 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 8

Preliminaries (VI) – Bayesian networks § Graphical models able to represent and efficiently manipulate n-dimensional probability distributions. § The knowledge obtained from a problem is encoded in a Belief network by means of the quantitative and qualitative componets: 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 9

Preliminaries (VII) – Bayesian networks • Qualitative part: Directed Acyclic Graph G=(V, E): 1. V (Nodes) Random variables, and 2. E (Arcs) (In)dependence relationships. • Quantitative part: A set of conditional distributions: 1. Drawn from the graph structure, 2. representing the strength of the relationships, 3. stored in each node. 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 10

MPEG-7 Video Retrieval Models based on Bayesian Networks (I) 4 Taking advantage of the structure of an MPEG-7 video: Video, Scenes, Shots, Frames 4 And of free text annotation tags in the. xml file… 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 11

MPEG-7 Video Retrieval Models based on Bayesian Networks (II) T 1 Sh 1 T 2 Sh 2 S 1 T 3 T 4 Sh 3 T 5 T 7 Sh 4 T 8 T 9 Sh 5 T 10 T 11 Sh 6 S 2 Sh 7 T 12 Sh 8 S 3 V 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 12

MPEG-7 Video Retrieval Models based on Bayesian Networks (III) Assesment of probability distributions: • Prior probability in term nodes: p(ti)=1/M. • Probability distributions in the rest of nodes: P(U | pa(U)). Problem: Great number of parents. Solution: Probability functions. 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 13

MPEG-7 Video Retrieval Models based on Bayesian Networks (IV) 1. Query term instantiation. 2. Run a propagation algorithm: p(u | Q), U. 3. Generate a ranking. Problem: • Great number of nodes in the graph. • Complex topology. Solution: Evaluation of probability functions in each layer. 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 14

MPEG-7 Video Retrieval Models based on Bayesian Networks (V) In shots: In Scenes and Videos: where vij and wij: Exact propagation 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 15

MPEG-7 Video Retrieval Models based on Bayesian Networks (VI) Once a relevance probability has been assigned to each unit, Which units are offered to the user? Those which present an accurate context, wider enaugh to be a good response to the query. How? Transforming the Bayesian Network into a Influence Diagram 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 16

MPEG-7 Video Retrieval Models based on Bayesian Networks (VII) Sh 4 Sh 5 D 1 Sh 6 D 3 D 2 U 1 S 4 S 3 D 5 D 4 U 4 27/02/2021 U 3 V 2 MPEG-7 video retrieval using Bayesian Networks U 5 17

MPEG-7 Video Retrieval Models based on Bayesian Networks (VIII) Integrated Tool: • Video capture from tv. • Automatic annotations form subtitles. • Manual annotations based on ontologies. • Querying and obtaining the best units. • Automatic generation of a video with the results. 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 18

MPEG-7 Video Retrieval Models based on Bayesian Networks (IX) Lalmas and Graves´ model: V S 1 Sh 1 S 2 Sh 2 Media. Information. DS Media. Profile. DS Sh 3 Sh 4 Sh 5 Sh 6 Creation. Information. DS Media. Quality. DS Media. Format. DS bbc C 2 C 3 C 4 C 5 C 7 C 8 C 9 C 10 C 11 dog Creation bbc dog and 27/02/2021 MPEG-7 video retrieval using Bayesian Networks 19

The end. . . Thank you very much MPEG-7 video retrieval using Bayesian Networks

- Slides: 20