Download presentation
Presentation is loading. Please wait.
1
Video retrieval using inference network A.Graves, M. Lalmas In Sig IR 02
2
Outline Background Mpeg 7 Inference network model Experiment Conclusion
3
Background CBVR: find the video upon demand The semantic gap: high level concepts and low level features Traditional method: retrieve by example Relevance feedback: tedious for video retrieval Retrieval by semantics?
4
Mpeg 7 Multimedia Content description interface Attach metadata to multimedia content Semantics: event, actor, place…… Structure: shot, scene, group…… Video as a structured document The information contained in the Mpeg 7 annotation can be exploited to perform semantic based video retrieval Mpeg 7 does not provide the solution to extract the annotation
5
Mpeg 7 Description definition language Descriptors Description schemes
6
Mpeg 7 Structured document and description Basil attempts to mend the car without success Basil Car Mend Carpark …… … …
7
Inference network Perform a ranking given many sources of evidence Document network (DN) Constructed from the document data, represents all the retrievable units Query network (QN) Constructed from the query, represents the information needs
8
Inference network The document network Document layer (retrievable units collection) Contextual layer (represents the contextual information about the document-concept links) Concept layer (represent all the concepts in the network)
9
Inference network Document network
10
Inference network Link weight calculation Structural: duration ratio (Between document nodes) Contextual (Between contextual nodes) sibling number, context size, frequency Context – concept (tf, idf)
11
Inference network Query network A framework of nodes that represents the information need Concept nodes Constraint operators (and, or, sum, not…), context-conceptual constraints
12
Inference network
13
Attachment and Evaluation Attachment: match the DN and QN, get a set of links between their nodes All the constraint satisfied Link inheritance: in the DN, document node can share the context nodes of its parent node Evaluation: calculated quantized similarity for each document node in the DN
14
Inference network Attachment Link the QN and DN such that: The concept nodes that contains same concepts (need a synonym dictionary) with weight 1 (firm) For constraint queries, adjust the weight
15
Inference network Attachment
16
Inference network Evaluation The evaluation process can be done toward all document nodes, it is calculated according to the query network Can be used in different applications: Retrieve scene, shot from one video Retrieve video from video collections …
17
Inference network Evaluation Back-propagate the QN-DN link weight back to each document nodes All the node will have a value Can derive a rank at different granularity level (Video, scene, shot)
18
Experiment 3 manually annotated video are employed as test data, totally 329 shots Annotations: for video shots and scenes Avg. precision: 69.25 Similarity order are as expected
19
Experiment The performance are quite dependent on the quality of the annotation data Efficient annotation methods will be quite helpful
20
Conclusion Propose a semantic video retrieval model based on Inference Network model, fully exploits the structural, conceptual, and contextual aspects of Mpeg 7. Parry the semantic gap problem
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.