Convex Quadratic Programming for Object Location Hao Jiang, Mark S. Drew and Ze-Nian Li School of Computing Science Simon Fraser University.

Slides:



Advertisements
Similar presentations
Algorithms for MAP estimation in Markov Random Fields Vladimir Kolmogorov University College London Tutorial at GDR (Optimisation Discrète, Graph Cuts.
Advertisements

Solving LP Models Improving Search Special Form of Improving Search
Principal Component Analysis Based on L1-Norm Maximization Nojun Kwak IEEE Transactions on Pattern Analysis and Machine Intelligence, 2008.
Graph Cut Algorithms for Computer Vision & Medical Imaging Ramin Zabih Computer Science & Radiology Cornell University Joint work with Y. Boykov, V. Kolmogorov,
Introduction to Markov Random Fields and Graph Cuts Simon Prince
Support Vector Machines
ICCV 2007 tutorial Part III Message-passing algorithms for energy minimization Vladimir Kolmogorov University College London.
Ziming Zhang*, Ze-Nian Li, Mark Drew School of Computing Science Simon Fraser University Vancouver, Canada {zza27, li, AdaMKL: A Novel.
Introduction to Belief Propagation and its Generalizations. Max Welling Donald Bren School of Information and Computer and Science University of California.
1 s-t Graph Cuts for Binary Energy Minimization  Now that we have an energy function, the big question is how do we minimize it? n Exhaustive search is.
Learning with Inference for Discrete Graphical Models Nikos Komodakis Pawan Kumar Nikos Paragios Ramin Zabih (presenter)
1 Minimum Ratio Contours For Meshes Andrew Clements Hao Zhang gruvi graphics + usability + visualization.
On Constrained Optimization Approach To Object Segmentation Chia Han, Xun Wang, Feng Gao, Zhigang Peng, Xiaokun Li, Lei He, William Wee Artificial Intelligence.
We propose a successive convex matching method to detect actions in videos. The proposed scheme does not need foreground/background separation, works in.
Yung-Lin Huang, Yi-Nung Liu, and Shao-Yi Chien Media IC and System Lab Graduate Institute of Networking and Multimedia National Taiwan University Signal.
Silhouettes in Multiview Stereo Ian Simon. Multiview Stereo Problem Input: – a collection of images of a rigid object (or scene) – camera parameters for.
1 Can this be generalized?  NP-hard for Potts model [K/BVZ 01]  Two main approaches 1. Exact solution [Ishikawa 03] Large graph, convex V (arbitrary.
Distributed Message Passing for Large Scale Graphical Models Alexander Schwing Tamir Hazan Marc Pollefeys Raquel Urtasun CVPR2011.
Last Time Pinhole camera model, projection
Jie Gao Joint work with Amitabh Basu*, Joseph Mitchell, Girishkumar Stony Brook Distributed Localization using Noisy Distance and Angle Information.
2010/5/171 Overview of graph cuts. 2010/5/172 Outline Introduction S-t Graph cuts Extension to multi-label problems Compare simulated annealing and alpha-
Human Posture Recognition with Convex Programming Hao Jiang, Ze-Nian Li and Mark S. Drew School of Computing Science Simon Fraser University Burnaby, BC,
A Combinatorial Maximum Cover Approach to 2D Translational Geometric Covering Karen Daniels, Arti Mathur, Roger Grinde University of Massachusetts Lowell.
Stereo & Iterative Graph-Cuts Alex Rav-Acha Vision Course Hebrew University.
Computing motion between images
Optical flow and Tracking CISC 649/849 Spring 2009 University of Delaware.
Job Scheduling Lecture 19: March 19. Job Scheduling: Unrelated Multiple Machines There are n jobs, each job has: a processing time p(i,j) (the time to.
Linear Solution to Scale and Rotation Invariant Object Matching Professor: 王聖智 教授 Student : 周 節.
Stereo Computation using Iterative Graph-Cuts
Comp 775: Graph Cuts and Continuous Maximal Flows Marc Niethammer, Stephen Pizer Department of Computer Science University of North Carolina, Chapel Hill.
EDA (CS286.5b) Day 19 Covering and Retiming. “Final” Like Assignment #1 –longer –more breadth –focus since assignment #2 –…but ideas are cummulative –open.
Branch and Bound Algorithm for Solving Integer Linear Programming
Computer vision: models, learning and inference
Stereo Matching & Energy Minimization Vision for Graphics CSE 590SS, Winter 2001 Richard Szeliski.
A Scale and Rotation Invariant Approach to Tracking Human Body Part Regions in Videos Yihang BoHao Jiang Institute of Automation, CAS Boston College.
A Rapid Stereo Matching Algorithm Based on Disparity Interpolation Gang Yao Yong Liu Bangjun Lei Dong Ren Institute of Intelligent Vision and Image Information.
Fast Approximate Energy Minimization via Graph Cuts
Finding a Hausdorff Core of a Polygon: On Convex Polygon Containment with Bounded Hausdorff Distance Reza Dorrigiv, Stephane Durocher, Arash Farzan, Robert.
Graph Cut 韋弘 2010/2/22. Outline Background Graph cut Ford–Fulkerson algorithm Application Extended reading.
Planar Cycle Covering Graphs for inference in MRFS The Typhon Algorithm A New Variational Approach to Ground State Computation in Binary Planar Markov.
Interactive Graph Cuts for Optimal Boundary & Region Segmentation of Objects in N-D Images (Fri) Young Ki Baik, Computer Vision Lab.
Object Stereo- Joint Stereo Matching and Object Segmentation Computer Vision and Pattern Recognition (CVPR), 2011 IEEE Conference on Michael Bleyer Vienna.
CS 4487/6587 Algorithms for Image Analysis
1 CPSC 320: Intermediate Algorithm Design and Analysis July 28, 2014.
Algorithms for MAP estimation in Markov Random Fields Vladimir Kolmogorov University College London.
1 Markov Random Fields with Efficient Approximations Yuri Boykov, Olga Veksler, Ramin Zabih Computer Science Department CORNELL UNIVERSITY.
Lecture 19: Solving the Correspondence Problem with Graph Cuts CAP 5415 Fall 2006.
Implicit Hitting Set Problems Richard M. Karp Erick Moreno Centeno DIMACS 20 th Anniversary.
1 Scale and Rotation Invariant Matching Using Linearly Augmented Tree Hao Jiang Boston College Tai-peng Tian, Stan Sclaroff Boston University.
Large-Scale Matrix Factorization with Missing Data under Additional Constraints Kaushik Mitra University of Maryland, College Park, MD Sameer Sheoreyy.
Resource Allocation in Hospital Networks Based on Green Cognitive Radios 王冉茵
Application: Multiresolution Curves Jyun-Ming Chen Spring 2001.
A global approach Finding correspondence between a pair of epipolar lines for all pixels simultaneously Local method: no guarantee we will have one to.
Tightening LP Relaxations for MAP using Message-Passing David Sontag Joint work with Talya Meltzer, Amir Globerson, Tommi Jaakkola, and Yair Weiss.
IE 312 Review 1. The Process 2 Problem Model Conclusions Problem Formulation Analysis.
Linear Solution to Scale and Rotation Invariant Object Matching Hao Jiang and Stella X. Yu Computer Science Department Boston College.
Iterative K-Means Algorithm Based on Fisher Discriminant UNIVERSITY OF JOENSUU DEPARTMENT OF COMPUTER SCIENCE JOENSUU, FINLAND Mantao Xu to be presented.
An O(bn 2 ) Time Algorithm for Optimal Buffer Insertion with b Buffer Types Authors: Zhuo Li and Weiping Shi Presenter: Sunil Khatri Department of Electrical.
Energy functions f(p)  {0,1} : Ising model Can solve fast with graph cuts V( ,  ) = T[  ] : Potts model NP-hard Closely related to Multiway Cut Problem.
Energy minimization Another global approach to improve quality of correspondences Assumption: disparities vary (mostly) smoothly Minimize energy function:
Water resources planning and management by use of generalized Benders decomposition method to solve large-scale MINLP problems By Prof. André A. Keller.
Lecture 5 Dynamic Programming
Markov Random Fields with Efficient Approximations
Object Matching Using a Locally Affine Invariant and Linear Programming Techniques - H. Li, X. Huang, L. He Ilchae Jung.
Efficient Graph Cut Optimization for Full CRFs with Quantized Edges
Chapter 6. Large Scale Optimization
Matteo Fischetti, University of Padova
Learning Incoherent Sparse and Low-Rank Patterns from Multiple Tasks
Chapter 6. Large Scale Optimization
Discrete Optimization
Presentation transcript:

Convex Quadratic Programming for Object Location Hao Jiang, Mark S. Drew and Ze-Nian Li School of Computing Science Simon Fraser University

2 Introduction  Object localization is an important task in computer vision Template

3 Object localization and labeling  Object localization can be formulated as labeling problems.  Consistent labeling  Find small cost label assignment.  Enforce labeling consistency of neighboring sites. p q fpfp fqfq c(p,f p ) c(q,f q ) TemplateTarget Object Site Label Neighboring relation (q-p) (f q -f p )

4 Previous methods for consistent labeling  Consistent labeling, in general form, is NP-hard.  Polynomial time schemes exist for special cases  Dynamic Programming (DP).  Max-flow [Ishikawa 2000, Roy 98].  Approximation schemes  Greedy (local searching) methods  Relaxation labeling (RL) [Rosenfeld 76].  Iterated Conditional Modes (ICM) [Besag 86]. Need good initialization and easily trapped in local minimum.  Global searching methods  Graduated Non-Convexity (GNC) [Blake & Zisserman 87]  Belief Propagation (BP) [Pearl 88, Weiss 2001].  Graph Cut (GC) [Boykov & Zabih 2001].

5 Focus of the research (# of labels in the order of hundred) (# of labels can be in the order of thousand or bigger) Many vision problems such as object matching, large scale motion, tracking etc will benefit from the solver. Well solved Hard to solve The focus of the research Consistent labeling  Previous methods become slow as the number of labels goes to the order of several thousand.

6 The trick of the proposed scheme  In stead of working on the original label space, we represent the label set with a small number basis labels.  We convert the hard problem into a sequence of simpler problems built using only the basis labels.  In this way, the size of the approximation problem is largely decoupled from the original label set.  Each sub-problem is a convex problem and can be globally optimized.  A successive relaxation implementation is used to zero in the target.

7 The non-linear optimization problem  The labeling problem can be solved by optimizing: p q fpfp fqfq c(p,f p ) c(q,f q ) (q-p) (f q -f p )

8 Convex relaxation  To convert the labeling cost term into linear functions, for each site, we define a basis label set.  Each label can then be represented as a linear combination of the basis labels.  The cost of the label is approximated by the linear combination of the costs of these basis labels. t c(s,t) J1J1 J2J2 fsfs f s = a * J 1 + b * J 2 c(s,f s ) = a * c(s,J 1 ) + b * c(s,J 2 ) a + b = 1

9 Convex relaxation (Cont’)  The L2 norm smoothness terms do not need additional conversions. j2j2 j3j3 j4j4 j5j5 j1j1 j 1  s,j 1 + j 2  s,j 2 + j 3  s,j 3 + j 4  s,j 4 + j 5  s,j 5 c(s,j 1 ) c(s,j 2 ) c(s,j 3 ) c(s,j 4 ) c(s,j 5 ) c(s,j 1 )  s,j 1 + c(s,j 2 )  s,j 2 + c(s,j 3 )  s,j 3 + c(s,j 4 )  s,j 4 + c(s,j 5 )  s,j 5  s,j 1 +  s,j 2 +  s,j 3 +  s,j 4 +  s,j 5 =1

10 Convex quadratic program (CQP) We have the convex quadratic program:

11 In some cases, the CQP is exact  If  are binary numbers, the mixed integer program is exactly equivalent to the original problem.  If c(s,t) is convex over t for each s 2 S, the CQP is equivalent to the continuous extension of the original problem. t t c(s,t) Feasible solution Continuous extension

12 In general it is an approximation  For general problems, the CQP approximates each labeling cost surface with the 3D lower convex hull. t =(x,y) c(s,t) The convexified surfaces are much simpler than the original ones. x y Label cost

13 The CQP can be greatly simplified  The most compact basis labels correspond to the lower convex hull vertices.  This shows a way to simplify the label set in labeling.  We can safely discard many labels without worrying about the problems met in previous methods. # of original label: 400 # of basis label: 20

14 Successive convexification To improve the approximation we use the successive convexification (SC) scheme as follows:

15 Experimental results Matching Random Dots

16 Experimental results (Cont’) Matching Random Dots

17 Experimental results (Cont’) Matching Leaf

18 Experimental results (Cont’) Matching Face

19 Experimental results (Cont’) Matching Hand

20 Conclusion  Set out an object localization method which can deal with textureless objects in strong background clutter.  Propose a successive convex quadratic method  CQP large decouples the number of target image points with the size of the convex program – It searches the whole target image quickly.  Successive convexification is studied to improve the result recursively.  Successive convexification is a general method can be applied to any convex regularization problem.

21 Thank you!

22 In summery  Label space is approximated with small number of basis labels.  Original problem is converted to a sequence of much easier convex programs and solved by successive relaxation.  The size of the convex program is largely decoupled with the number of candidate labels Convexification Trust Region Shrinking Template Target Image Matching Costs Lower Convex Hull