Download presentation
Presentation is loading. Please wait.
1
Digital Video Library - Jacky Ma
2
Introduction Overview Application System Design Technologies
Speech Recognition Image Understanding Metadata Extraction Challenges
3
Overview Make large video library to be searchable information resources Video Captures the experience of society News, TV, Movie…etc Search and Discovery Automated knowledge extraction from video Integration of speech, image, and natural language understanding for library creation and exploration
4
Application Areas Education and training
Consumer and business access to news and information of interest Entertainment Interactive television Meeting/corporate memory Video conferences
5
Application of Diverse Technologies
Image Understanding Scene Understanding Speech Recognition Metadata/Entity Extraction Natural Language Processing More… Database, Network, User Interface...
6
Library Creation
7
Library Exploration
8
Information Retrieval
Given a large collection of multimedia records, find similar/interesting things Allow fast, approximate queries Find rules/patterns Similarity search Find pairs of documents that are similar Find medical cases similar to Smith’s Find pairs of stocks that move in sync
9
Indexing for Multimedia
Speech Recognition Functions Generates transcript to enable text-base retrieval Provides speech interface to digital library Supplies necessary information for library segmentation and multimedia abstraction
10
Speech Recognition
11
Speech Recognition - cont’d
12
Speech Recognition - cont’d
13
Indexing for Multimedia
Image Understanding Functions Scene segmentation Similarity matching Camera motion determination and object tracking OCR on video text and titles Face detection and recognition Future: Object identification, scene characterization
14
Image Understanding Dimensions of matching are color, texture, shape
Methods Color histogram, Region-based analysis But “similar” has different meaning for different people, or same person but different situation
15
Image Understanding - cont’d
16
Indexing for Multimedia
Metadata Extraction Higher level of abstract knowledge Story summaries Real world knowledge people location time event...
17
Metadata Extraction
18
Challenges Multilingual Processing Cognitive Processing
Library Interoperability Intellectual Property Security Issues
19
Thank you
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.