Efficient Video Browsing (Ch. 1)
By Teklu
Q1. Discuss the three levels of browsing .How multiple synchronized views enable to achieve them.
Q2. Discuss Shot Boundary Detection Technique. Explain why it is important in creating multiple Synchronized View to video contents.
Q3. What is Time Scale Modification? How it is different from Adaptive Fast Playback
By 서석준
Q1. For the way this paper suggests to streaming synchronized views, what is the pros and cons of that way?
Q2. Why the movieDNA is needed?
Q3. How the speed affects user's comprehension?
Temporal Video Boundaries (Ch. 3)
By 김경민
Q1. When does segmentation by megaboundariesoccure? Give an example.
Q2. When deciding a microboundary does existence of memory has a difference?
Q3. When deciding macro boundary, what is single descent with secondary voting attributes?
By Margad
Q1.What is the advantage of using GA in our commercial detection problem ?
Q2.What is trigger verifier model and how are we using it ?
Q3.What can we achieve by doing temporal video boundaries detection ?
Movie Content Analysis, Indexing and Skimming (Ch. 5)
By 김덕주
Q1. What is the objective of content-based video analysis?
Q2. Why supervised identification has limitation?
Q3. Why should use integrated media data?
Face Recognition and Retrieval in Video
By 오준혁
Q1. What is the advantage of video over single image in face recognition and why?
Q2. What are three approaches for face recognition in video and what is the difference between them?
Q3. What is the relationship between (a) face detection and tracking (b) face recognition (c) face retrieval?
Understanding the Semantics of Media (Ch. 8)
By Camilo
Q1. What kind of application does SVD has? How is it used in this paper?
Q2. What does MPESAR stands for? What does this system do?
Q3. How does MPESAR generally works?
A Human-Centered Computing Framework to Enable Personalized News Video Recommendation
By 오준혁
Q1) How to detect news topic from video?
Q2) How to measure inter-topic association?
(hint :Keyword association, Co-occurrence)
Q3) How to measure interestingness of news topic?
(hint :Popularity, Importance)
By Camilo
missing
Statistical Techniques for Video Analysis and Searching (Ch. 9)
By Anton
Q1. What are the main approaches for video analysis based on vector model indexing and interactive search fusion?
Q2. Which technique we apply in this approaches?
Q3. What detectors we can use for that technique?
By Margad
Q1. What is the difference between manual and interactive video search fusion?
Q2. What is CBR and why is it good for video searching ?
Q3. What is multi example search and what is the advantage of using it ?
Unsupervised Mining of Statistical Temporal Structures in Video (Ch. 10)
By 강현수
Q1. What is supervised learning?
Q2. What is the benefit of using unsupervised learning?
Q3. Into what aspects does the feature selection can divided and why?
Q4. Why do we need to adopt randomized search strategies?
Pseudo-Relevance Feedback for Multimedia Retrieval (Ch. 11)
By 서석준
Q1. How does this paper approach to content-based video retrieval?
Q2. What is the advantage of negative PRF?
Q3. What doesthis paper do to remove extreme outliers?
Video Repeat Recognition and Mining by Visual Features
By 강현수
Q1.What are the applications that video repeat recognition can be used?
Q2. What are the techniques used for the algorithm on the known and unknown video repeat recognition, and why?
(hint :known video, unknown video)
Q3. What are the main issues in unknown video repeat mining and what are the methods used?
Q4. Assume there were 25 kinds of repeat clips with totally 200 instances in news A, while 50 kinds with totally 500 instances in news B. In a test, for news A our trained detector showed 80% of recall with the total detection of 320 instances, and for news B, our trained detector showed 85% of precision with the total detection of 400 instances. (a) What is precision of news A and recall of news B? (b) What are precision and recall of news A and B combined?