Transformation of an Uncertain Video Search Pipeline to a Sketch-Based Visual Analytics Loop Philip A. Legg 1,2, David H.S. Chung 2, Matthew L. Parry 2,

Slides:



Advertisements
Similar presentations
Pseudo-Relevance Feedback For Multimedia Retrieval By Rong Yan, Alexander G. and Rong Jin Mwangi S. Kariuki
Advertisements

SPONSORED BY SA2014.SIGGRAPH.ORG Annotating RGBD Images of Indoor Scenes Yu-Shiang Wong and Hung-Kuo Chu National Tsing Hua University CGV LAB.
Visual Analytics Research at WPI Dr. Matthew Ward and Dr. Elke Rundensteiner Computer Science Department.
Self Organizing Maps. This presentation is based on: SOM’s are invented by Teuvo Kohonen. They represent multidimensional.
ADVISE: Advanced Digital Video Information Segmentation Engine
System Design and Analysis
CS335 Principles of Multimedia Systems Content Based Media Retrieval Hao Jiang Computer Science Department Boston College Dec. 4, 2007.
Multimedia Search and Retrieval Presented by: Reza Aghaee For Multimedia Course(CMPT820) Simon Fraser University March.2005 Shih-Fu Chang, Qian Huang,
Retrieval Evaluation. Brief Review Evaluation of implementations in computer science often is in terms of time and space complexity. With large document.
T.Sharon 1 Internet Resources Discovery (IRD) Introduction to MMIR.
Visual Information Retrieval Chapter 1 Introduction Alberto Del Bimbo Dipartimento di Sistemi e Informatica Universita di Firenze Firenze, Italy.
MULTIPLE MOVING OBJECTS TRACKING FOR VIDEO SURVEILLANCE SYSTEMS.
Using 3D-SURFER. Before you start 3D-Surfer can be accessed at For visualization.
The Project AH Computing. Functional Requirements  What the product must do!  Examples attractive welcome screen all options available as clickable.
MarketLine HQ ADVANTAGE – your subscription service Explore today at
MULTIMEDIA DEVELOPMENT
An Interactive Virtual Endoscopy Tool With Automated Path Generation Delphine Nain, MIT AI Laboratory. Thesis Advisor : W. Eric. L Grimson, MIT AI Laboratory.
EE 492 ENGINEERING PROJECT LIP TRACKING Yusuf Ziya Işık & Ashat Turlibayev Yusuf Ziya Işık & Ashat Turlibayev Advisor: Prof. Dr. Bülent Sankur Advisor:
TAUCHI – Tampere Unit for Computer-Human Interaction Visualizing gaze path for analysis Oleg Špakov MUMIN workshop 2002, Tampere.
An Internet of Things: People, Processes, and Products in the Spotfire Cloud Library Dr. Brand Niemann Director and Senior Data Scientist/Data Journalist.
1 Multiple Classifier Based on Fuzzy C-Means for a Flower Image Retrieval Keita Fukuda, Tetsuya Takiguchi, Yasuo Ariki Graduate School of Engineering,
An MPEG-7 Based Content- aware Album System for Consumer Photographs 2003/12/18 Chen-Hsiu Huang, Chih-Hao Shen, Chun-Hsiang Huang and Ja-Ling Wu Communication.
A Novel Local Patch Framework for Fixing Supervised Learning Models Yilei Wang 1, Bingzheng Wei 2, Jun Yan 2, Yang Hu 2, Zhi-Hong Deng 1, Zheng Chen 2.
Creating Graphical User Interfaces (GUI’s) with MATLAB By Jeffrey A. Webb OSU Gateway Coalition Member.
Quantitative Analyses of Human Pubic Symphyseal Morphology Using Three Dimensional Data: The Potential Utility for Aging Adult Human Skeletons Matthew.
Non-Photorealistic Rendering and Content- Based Image Retrieval Yuan-Hao Lai Pacific Graphics (2003)
INTERACTIVELY BROWSING LARGE IMAGE DATABASES Ronald Richter, Mathias Eitz and Marc Alexa.
BLAST Slides adapted & edited from a set by Cheryl A. Kerfeld (UC Berkeley/JGI) & Kathleen M. Scott (U South Florida) Kerfeld CA, Scott KM (2011) Using.
2005/12/021 Fast Image Retrieval Using Low Frequency DCT Coefficients Dept. of Computer Engineering Tatung University Presenter: Yo-Ping Huang ( 黃有評 )
MOVIE RETRIEVAL SYSTEM INFORMATION VISUALIZATION & PROPOSING NEW INTERFACE IAT 814 Adrian Bisek.
Chapter 3 Response Charts.
Tight Coupling of Dynamic Query Filters with Starfield Displays / Spotfire.net Desktop By Chris Ahlberg and Ben Shneiderman / Spotfire Inc. IC280 5/9/02.
David Adams ATLAS DIAL: Distributed Interactive Analysis of Large datasets David Adams BNL August 5, 2002 BNL OMEGA talk.
By: David Gelbendorf, Hila Ben-Moshe Supervisor : Alon Zvirin
User Interface Design Patterns: Part 1 Kirsten McCane.
2006/10/25 1 A Virtual Endoscopy System Author : Author : Anna Vilanova 、 Andreas K ö nig 、 Eduard Gr ö ller Source :Machine Graphics and Vision, 8(3),
Anomaly Detection in GPS Data Based on Visual Analytics Kyung Min Su - Zicheng Liao, Yizhou Yu, and Baoquan Chen, Anomaly Detection in GPS Data Based on.
Query by Image and Video Content: The QBIC System M. Flickner et al. IEEE Computer Special Issue on Content-Based Retrieval Vol. 28, No. 9, September 1995.
Jianping Fan Department of Computer Science University of North Carolina at Charlotte Charlotte, NC Relevance Feedback for Image Retrieval.
IntroGoalCrowdPredictionWrap-up 1/26 Learning Debugging and Hacking the User Remco Chang Assistant Professor Tufts University.
White Pages Team Grey Pages Facilitator Team & Facilitator Guide for School-wide Reading Leadership Team Meetings Elementary.
BRAIN SCAN  Brain scan is an interactive quiz for use as a revision/ learning reinforcement tool that accompanies the theory package.  To answer a question.
How to use C OBI T implementation resources Brian Selby Director of C OBI T Initiatives ISACA.
A Generic Toolkit for Electronic Editions of Medieval Manuscripts
Visual Information Retrieval
Visualization of Eye Gaze Data using Heat Maps
Data visualization for Decision Making
C2A: Crowd Consensus Analytics for Virtual Colonoscopy
Getting Started with Power Query
Project Overview Introduction to Factory Automation Numerical Control
Webinar – New KStutor Overview 25th October 2013
Planning for Institutional Effectiveness Manager PIE Planning & Resources 2018 User Guide.
VP PIE Planning & Resources
What can the Altmetric Explorer help you with?
N. Capp, E. Krome, I. Obeid and J. Picone
Planning for Institutional Effectiveness PIE Planning & Resources
Can Sorting Theory Contribute to Visual Analytics Theory?
Test Information Distribution Engine (TIDE) Training
E190Q – Final Project Presenation
Distributed Probabilistic Range-Aggregate Query on Uncertain Data
Visual Analytics for Big Video Visualization
Ying Dai Faculty of software and information science,
Microtubule Structure at 8 Å Resolution
Microtubule Structure at 8 Å Resolution
EE 492 ENGINEERING PROJECT
BLAST Slides adapted & edited from a set by
BLAST Slides adapted & edited from a set by
What is a System? A system is a collection of interrelated components that work together to perform a specific task.
Jiwon Kim Steve Seitz Maneesh Agrawala
Data Analytics Case Study
Presentation transcript:

Transformation of an Uncertain Video Search Pipeline to a Sketch-Based Visual Analytics Loop Philip A. Legg 1,2, David H.S. Chung 2, Matthew L. Parry 2, Rhodri Bown 3, Mark W. Jones 2, Iwan W. Griffiths 2 and Min Chen 1. University of Oxford 1, Swansea University 2, Welsh Rugby Union 3

Sports Analysis – What is Rugby?

Problem Sports analysts currently study hours of video content to find clips of action for a varierty of needs… Team and Player de-briefings Team committee meetings Match highlights Coaching tactical analysis How can we support this process???

Typical Machine Learning Pipeline Annotated training data provided by the user System learns a model that supports the trained input data Unseen test data is provided by the user System applies model to new test data to obtain a resulting output

Annotated training data provided by the user System learns a model that supports the trained input data Unseen test data is provided by the user System applies model to new test data to obtain a resulting output Typical Machine Learning Pipeline Analyst do not have the time or the technical ability to manually annotate training data! What if the analyst’s requirements change? Trained model becomes useless for their task!

Typical Machine Learning Pipeline

Visual Analytics Loop User inputs their search query into the system. System compares input against current model and returns results. User accepts / rejects results returned by the system. Model is refined based on the user’s acceptance of the results.

Visual Analytics Loop User inputs their search query into the system. System compares input against current model and returns results. User accepts / rejects results returned by the system. Model is refined based on the user’s acceptance of the results.

Interface Overview Sketch Input Rejected Results Model Vis. Accepted Results Search Space Vis. Search Results

User Input for searching Rugby video Sketch-based input. Intuitive drawing tools used to define: Movement paths (red) Regions of interest (blue) Distance range (green)

Obtaining data from Rugby video Top-down view to obtain full pitch coverage. Background filtering to extract players. Team centroid calculated from player positions. Convex hull used to bound team within shape.

Visual Analytics Loop User inputs their search query into the system. System compares input against current model and returns results. User accepts / rejects results returned by the system. Model is refined based on the user’s acceptance of the results.

Interface Overview Sketch Input Rejected Results Model Vis. Accepted Results Search Space Vis. Search Results

Search Space Visualization Timeline view for browsing video. Action glyphs to summarize video content. Plot shows the overall similarity score from the analytical search model.

Interface Overview Sketch Input Rejected Results Model Vis. Accepted Results Search Space Vis. Search Results

Model Visualization Parallel co-ordinates view. Each polyline corresponds to a video segment. Value for each axis show how each video segment scores against that particular metric within the analytical search model.

Model Visualization Pie slice above each axis shows the weighted contribution that metric provides to the overall similarity score. Weights can be adjusted through acceptance / rejection of results, or manual over-ride.

Visual Analytics Loop User inputs their search query into the system. System compares input against current model and returns results. User accepts / rejects results returned by the system. Model is refined based on the user’s acceptance of the results.

Interface Overview Sketch Input Rejected Results Model Vis. Accepted Results Search Space Vis. Search Results

Model Learning Thumbnail selection highlights corresponding polyline in model visualization. Normalized weights are updated based on user selection. Acceptance: w n = w n (1 + ω) Rejection: w n = w n (1 - ω)

Thumbnail Viewer To further guide user judgement, thumbnail can be interactively explored. Thumbnail visualization Annotated top-down video showing distances and team groupings Standard match views

Visual Analytics Loop User inputs their search query into the system. System compares input against current model and returns results. User accepts / rejects results returned by the system. Model is refined based on the user’s acceptance of the results.

Video

Summary We present a novel approach to searching large video data using a visual analytics loop. Model supports learning from user input to refine results – less time- consuming than preparing training data for machine learning. Visual analytics allow the user to assess the performance of the analytical search model. Visual analytics allow the user to search and examine match video content in greater detail.

Thank you for your attention. Philip A. Legg 1,2, David H.S. Chung 2, Matthew L. Parry 2, Rhodri Bown 3, Mark W. Jones 2, Iwan W. Griffiths 2 and Min Chen 1. University of Oxford 1, Swansea University 2, Welsh Rugby Union 3