Leveraging High-Level Expectations for Activity Recognition

Slides:



Advertisements
Similar presentations
KAIS T The Vision of Autonomic Computing Jeffrey O. Kephart, David M Chess IBM Watson research Center IEEE Computer, Jan 발표자 : 이승학.
Advertisements

1 Approximated tracking of multiple non-rigid objects using adaptive quantization and resampling techniques. J. M. Sotoca 1, F.J. Ferri 1, J. Gutierrez.
Tracking Objects with Dynamics Computer Vision CS 543 / ECE 549 University of Illinois Derek Hoiem 04/21/15 some slides from Amin Sadeghi, Lana Lazebnik,
A Bayesian algorithm for tracking multiple moving objects in outdoor surveillance video Department of Electrical Engineering and Computer Science The University.
Zhu, Song-Chun and Mumford, David. A Stochastic Grammar of Images. Foundations and Trends in Computer Graphics and Vision 2(4), (2006) Hemerson.
Process Query Systems ENGS 112 Lecture 7. Process Query Systems (PQS) vs Data Base Systems (DBS) Data Base System Process Query System Data Sources Data.
Columbia Hypermedia IMmersion Environment CHIME.
Hierarchical GUI Test Case Generation Using Automated Planning Atif M. Memon, Student Member, IEEE, Martha E. Pollack, and Mary Lou Soffa, Member, IEEE.
Tracking Video Objects in Cluttered Background
Formal verification Marco A. Peña Universitat Politècnica de Catalunya.
1 Video Surveillance systems for Traffic Monitoring Simeon Indupalli.
User Interface Design Principles Gabriel Spitz 1 Lecture # 4.
Learning Table Extraction from Examples Ashwin Tengli, Yiming Yang and Nian Li Ma School of Computer Science Carnegie Mellon University Coling 04.
User Characteristics & Design Principles Gabriel Spitz 1 Lecture # 11.
[ §6 : 1 ] 6. Basic Methods II Overview 6.1 Models 6.2 Taxonomy 6.3 Finite State Model 6.4 State Transition Model 6.5 Dataflow Model 6.6 User Manual.
Activity Diagram. C-S 5462 Activity diagram a diagram that represents a sequence of activities similar to flow chart, used in 1960’s and 1970’s an activity.
Chapter 10 State Machine Diagrams
Visual Object Tracking Xu Yan Quantitative Imaging Laboratory 1 Xu Yan Advisor: Shishir K. Shah Quantitative Imaging Laboratory Computer Science Department.
Probabilistic Context Free Grammars for Representing Action Song Mao November 14, 2000.
1 ECE 453 – CS 447 – SE 465 Software Testing & Quality Assurance Instructor Kostas Kontogiannis.
K. J. O’Hara AMRS: Behavior Recognition and Opponent Modeling Oct Behavior Recognition and Opponent Modeling in Autonomous Multi-Robot Systems.
ISBN Chapter 3 Describing Semantics -Attribute Grammars -Dynamic Semantics.
Tracking People by Learning Their Appearance Deva Ramanan David A. Forsuth Andrew Zisserman.
Xusheng Xiao North Carolina State University CSC 720 Project Presentation 1.
Human Activity Recognition at Mid and Near Range Ram Nevatia University of Southern California Based on work of several collaborators: F. Lv, P. Natarajan,
Multiple track hypotheses maintained for each possible blob-object mapping Parse driven by blob interaction events Domain-general events are detected (e.g.,
Gabriel Spitz 1. Light Switch Issues  Consistency – The switch design is inconsistent with common light switches  Visibility – No visible cue regarding.
Lucent Technologies - Proprietary 1 Interactive Pattern Discovery with Mirage Mirage uses exploratory visualization, intuitive graphical operations to.
Stochastic Grammars: Overview Representation: Stochastic grammar Representation: Stochastic grammar Terminals: object interactions Terminals: object interactions.
Using DataShop Tools to Model Students Learning Statistics Marsha C. Lovett Eberly Center & Psychology Acknowledgements to: Judy Brooks, Ken Koedinger,
Visual Recognition Tutorial1 Markov models Hidden Markov models Forward/Backward algorithm Viterbi algorithm Baum-Welch estimation algorithm Hidden.
1 Minimum Bayes-risk Methods in Automatic Speech Recognition Vaibhava Geol And William Byrne IBM ; Johns Hopkins University 2003 by CRC Press LLC 2005/4/26.
An Ontology framework for Knowledge-Assisted Semantic Video Analysis and Annotation Centre for Research and Technology Hellas/ Informatics and Telematics.
Wolfgang Runte Slide University of Osnabrueck, Software Engineering Research Group Wolfgang Runte Software Engineering Research Group Institute.
The Role of Simulation in the 2010 Highway Capacity Manual ITE Western District Annual Meeting June 28, 2010 Loren Bloomberg/CH2M HILL Santa Ana, CA
Modelling and Solving Configuration Problems on Business
Constraint Satisfaction Problems (Chapter 6)
Agenda Preliminaries Motivation and Research questions Exploring GLL
SNS College of Engineering Department of Computer Science and Engineering AI Planning Presented By S.Yamuna AP/CSE 5/23/2018 AI.
Introduction to Parsing (adapted from CS 164 at Berkeley)
Observer Design Pattern
Conditional Random Fields for ASR
Potter’s Wheel: An Interactive Data Cleaning System
International Research and Development Institute Uyo
Datamining : Refers to extracting or mining knowledge from large amounts of data Applications : Market Analysis Fraud Detection Customer Retention Production.
Software engineering USER INTERFACE DESIGN.
Gabor Madl Ph.D. Candidate, UC Irvine Advisor: Nikil Dutt
Above and below the object level
The Extensible Tool-chain for Evaluation of Architectural Models
Identifying Human-Object Interaction in Range and Video Data
Chapter 1 Database Systems
File Systems and Databases
An Evolutional Model for Operation-driven Visualization Design
Leveraging High-Level Expectations for Activity Recognition
The Vision of Autonomic Computing
A Hierarchical Bayesian Look at Some Debates in Category Learning
Automated Software Integration
Model Comparison: A Key Challenge for Transformation Testing and Version Control in Model Driven Software Development Yuehua Lin, Jing Zhang, Jeff Gray.
Chapter 1 Database Systems
Parsing Unrestricted Text
Detection and tracking of individual SNAP-tagged proteins on the surface of living cells. Detection and tracking of individual SNAP-tagged proteins on.
Med-Fi Prototype Presentation
Design Yaodong Bi.
Getting started – Example 1
Some Programming Paradigms
Some Important Skills Every Software Testers Should Have
Range Likelihood Tree: A Compact and Effective Representation for Visual Exploration of Uncertain Data Sets Ohio State University (Shen) Problem: Uncertainty.
ACM programming contest
Interaction Models (2): Activity Diagrams
Design.
Presentation transcript:

Leveraging High-Level Expectations for Activity Recognition David Minnen, Irfan Essa, Thad Starner GVU Center / Georgia Institute of Technology {dminn,irfan,thad}@cc.gatech.edu http://www.cc.gatech.edu/cpl/projects/expectationGrammars Automatically detect and decompose an activity from video Activity manually specified as a parameterized stochastic grammar Feedback from high-level parse influences interpretation of object identities Experiments use the Towers of Hanoi task to focus on activity recognition rather than visual appearance modeling Overview & Goals ToH -> Setup, enter(hand), Solve, exit(hand); Setup -> TowerPlaced, exit(hand); TowerPlaced -> enter(hand, block_A, block_B, block_C), Put_1(block_A, block_B, block_C); Solve -> state(Tower = TowerStart), MakeMoves, state(Tower = TowerGoal); MakeMoves -> Move(block) [0.1] | Move(block), MakeMoves [0.9]; Move -> Move_1-2 | Move_1-3 | Move_2-1 | Move_2-3 | Move_3-1 | Move_3-2; Move_1-2 -> Grab_1, Put_2; Move_1-3 -> Grab_1, Put_3; Move_2-1 -> Grab_2, Put_1; Move_2-3 -> Grab_2, Put_3; Move_3-1 -> Grab_3, Put_1; Move_3-2 -> Grab_3, Put_2; Grab_1 -> touch_1, remove_1(hand, ~) | touch_1(~), remove_last_1(~); Grab_2 -> touch_2, remove_2(hand, ~) | touch_2(~), remove_last_2(~); Grab_3 -> touch_3, remove_3(hand, ~) | touch_3(~), remove_last_3(~); Put_1 -> release_1(~) | touch_1, release_1; Put_2 -> release_2(~) | touch_2, release_2; Put_3 -> release_3(~) | touch_3, release_3; Stochastic Grammar for Towers of Hanoi Task Approach Multiple track hypotheses maintained for each possible blob-object mapping Parse driven by blob interaction events Domain-general events are detected (e.g., merge, split, enter, exit) Events transformed into activity- specific terminals of the grammar Abstract model used to detect and prune semantically inconsistent interpretations Likelihood of parse computed from production and symbol probabilities Activity recognized by selecting the most likely, semantically consistent hypothesis Data Flow Identifying Objects by Action Instead of Appearance Recognition of Towers of Hanoi with Occlusion [113-739 (1-29) ToH . ] [113-739 (1-29) Setup, Solve, exit(hand) . ] [113-234 (1-4) TowerPlaced, exit(hand), enter(hand) . ] [113-169 (1-2) enter(hand, red, green, blue), Put_1(red, green, blue) . ] [169-169 (2-2) release_1 . ] [248-733 (5-28) state(Tower=TowerStart), MakeMoves, state(Tower=TowerGoal) . ] [248-733 (5-28) Move(block), MakeMoves . ] [248-315 (5-7) Move_1-3 . ] [248-315 (5-7) Grab_1, Put_3 . ] [248-270 (5-6) touch_1, remove_1(hand) . ] [315-315 (7-7) release_3 . ] [337-733 (8-28) Move(block), MakeMoves . ] [337-385 (8-10) Move_1-2 . ] [337-385 (8-10) Grab_1, Put_2 . ] [337-350 (8-9) touch_1, remove_1(hand) . ] [385-385 (10-10) release_2 . ] … Partial Parse of ToH Task Limitations & Future Work Difficult to represent complex temporal constraints such as concurrency and quantitative temporal relationships Inefficient deletion error recovery (exponential growth) Requires activities with strong temporal relationships to disambiguate object identities and actions without relying on appearance models Comparison with other temporal models (e.g., n-gram models, HMMs, DBNs)