Discussion Class 10 Informedia 1 Discussion Classes Format

Discussion Class 10 Informedia 1

Discussion Classes Format: Question Ask a member of the class to answer. Provide opportunity for others to comment. When answering: Stand up. Give your name. Make sure that the TA hears it. Speak clearly so that all the class can hear. Suggestions: Do not be shy at presenting partial answers. Differing viewpoints are welcome. 2

Question 1: Automated Extract of Text Informedia extracts text automatically from video in three different ways. (a) What are they? (b) How effective would you expect each to be for information retrieval? 3

Question 2: Speech recognition and retrieval performance What does the following graph tell us? 4

Question 3: Browsing (a) Informedia provides multiple levels of abstraction and summarization. Why? (b) Describe the function of each of the following: i ii iv visual icons one-line headlines film strip views video skims v transcript following of audio track (c) How is each produced? 5

Question 4: Content and Context "The Informedia processing provided state of the art access to video by content. Current efforts will communicate information trends across time, space, and sources by furthering analysis and understanding of the context as well as the content. " (a) What does the word content mean in this paragraph? (b) What does the word context mean in this paragraph? 6

Question 5: Image Understanding What role does image understanding perform in each of the following in Informedia? (a) scene break detection (icon selection, film-strip and skimming, segmentation) (b) image similarity matching (c) camera motion and object tracking (d) video-OCR (e) face detection and association 7

Question 6: Scalability Throughout the article there are mentions of scale. (a) What has the project done to evaluate scalability? (b) What are the weaknesses that might prove difficult on a larger scale? 8

Question 7: Overall Informedia had the following objectives: (a) retrieval performance in the presence of inaccuracy and ambiguity (b) approximate match in meaning and visualization (c) presentation and reuse of video content as a new data type with space and time constraints (d) interoperability in the presence of restricted use intellectual property and the absence of data and protocol standards. How well has each been achieved? 9