Results 1 
9 of
9
An efficient branchandbound algorithm for optimal human pose estimation
"... Human pose estimation in a static image is a challenging problem in computer vision in that body part configurations are often subject to severe deformations and occlusions. Moreover, efficient pose estimation is often a desirable requirement in many applications. The tradeoff between accuracy an ..."
Abstract

Cited by 20 (3 self)
 Add to MetaCart
(Show Context)
Human pose estimation in a static image is a challenging problem in computer vision in that body part configurations are often subject to severe deformations and occlusions. Moreover, efficient pose estimation is often a desirable requirement in many applications. The tradeoff between accuracy and efficiency has been explored in a large number of approaches. On the one hand, models with simple representations (like tree or star models) can be efficiently applied in pose estimation problems. However, these models are often prone to body part misclassification errors. On the other hand, models with rich representations (i.e., loopy graphical models) are theoretically more robust, but their inference complexity may increase dramatically. In this work, we propose an efficient and exact inference algorithm based on branchandbound to solve the human pose estimation problem on loopy graphical models. We show that our method is empirically much faster (about 74 times) than the stateoftheart exact inference algorithm [21]. By extending a stateoftheart tree model [16] to a loopy graphical model, we show that the estimation accuracy improves for most of the body parts (especially lower arms) on popular datasets such as Buffy [7] and Stickmen [5] datasets. Finally, our method can be used to exactly solve most of the inference problems on Stretchable Models [18] (which contains a few hundreds of variables) in just a few minutes.
Towards Efficient and Exact MAPInference for Large Scale Discrete Computer Vision Problems via Combinatorial Optimization
, 2013
"... Discrete graphical models (also known as discrete Markov random fields) are a major conceptual tool to model the structure of optimization problems in computer vision. While in the last decade research has focused on fast approximative methods, algorithms that provide globally optimal solutions have ..."
Abstract

Cited by 9 (1 self)
 Add to MetaCart
Discrete graphical models (also known as discrete Markov random fields) are a major conceptual tool to model the structure of optimization problems in computer vision. While in the last decade research has focused on fast approximative methods, algorithms that provide globally optimal solutions have come more into the research focus in the last years. However, large scale computer vision problems seemed to be out of reach for such methods. In this paper we introduce a promising way to bridge this gap based on partial optimality and structural properties of the underlying problem factorization. Combining these preprocessing steps, we are able to solve grids of size 2048×2048 in less than 90 seconds. On the hitherto unsolvable Chinese character dataset of Nowozin et al. we obtain provably optimal results in 56 % of the instances and achieve competitive runtimes on other recent benchmark problems. While in the present work only generalized Potts models are considered, an extension to general graphical models seems to be feasible.
Branch and bound algorithm for dependency parsing with nonlocal features
 TACL
, 2013
"... Graph based dependency parsing is inefficient when handling nonlocal features due to high computational complexity of inference. In this paper, we proposed an exact and efficient decoding algorithm based on the Branch and Bound (B&B) framework where nonlocal features are bounded by a linear com ..."
Abstract

Cited by 4 (1 self)
 Add to MetaCart
Graph based dependency parsing is inefficient when handling nonlocal features due to high computational complexity of inference. In this paper, we proposed an exact and efficient decoding algorithm based on the Branch and Bound (B&B) framework where nonlocal features are bounded by a linear combination of local features. Dynamic programming is used to search the upper bound. Experiments are conducted on English PTB and Chinese CTB datasets. We achieved competitive Unlabeled Attachment Score (UAS) when no additional resources are available: 93.17% for English and 87.25 % for Chinese. Parsing speed is 177 words per second for English and 97 words per second for Chinese. Our algorithm is general and can be adapted to nonprojective dependency parsing or other graphical models.
C.: Global MAPoptimality by shrinking the combinatorial search area with convex relaxation
, 2013
"... We consider energy minimization for undirected graphical models, also known as the MAPinference problem for Markov random fields. Although combinatorial methods, which return a provably optimal integral solution of the problem, made a significant progress in the past decade, they are still typicall ..."
Abstract

Cited by 4 (2 self)
 Add to MetaCart
(Show Context)
We consider energy minimization for undirected graphical models, also known as the MAPinference problem for Markov random fields. Although combinatorial methods, which return a provably optimal integral solution of the problem, made a significant progress in the past decade, they are still typically unable to cope with largescale datasets. On the other hand, large scale datasets are often defined on sparse graphs and convex relaxation methods, such as linear programming relaxations then provide good approximations to integral solutions. We propose a novel method of combining combinatorial and convex programming techniques to obtain a global solution of the initial combinatorial problem. Based on the information obtained from the solution of the convex relaxation, our method confines application of the combinatorial solver to a small fraction of the initial graphical model, which allows to optimally solve much larger problems. We demonstrate the efficacy of our approach on a computer vision energy minimization benchmark. 1
A Fast and Exact Energy Minimization Algorithm for Cycle MRFs
"... The presence of cycles gives rise to the difficulty in performing inference for MRFs. Handling cycles efficiently would greatly enhance our ability to tackle general MRFs. In particular, for dual decomposition of energy minimization (MAP inference), using cycle subproblems leads to a much tighter re ..."
Abstract

Cited by 2 (1 self)
 Add to MetaCart
(Show Context)
The presence of cycles gives rise to the difficulty in performing inference for MRFs. Handling cycles efficiently would greatly enhance our ability to tackle general MRFs. In particular, for dual decomposition of energy minimization (MAP inference), using cycle subproblems leads to a much tighter relaxation than using trees, but solving the cycle subproblems turns out to be the bottleneck. In this paper, we present a fast and exact algorithm for energy minimization in cycle MRFs, which can be used as a subroutine in tackling general MRFs. Our method builds on junctiontree message passing, with a large portion of the message entries pruned for efficiency. The pruning conditions fully exploit the structure of a cycle. Experimental results show that our algorithm is more than an order of magnitude faster than other stateoftheart fast inference methods, and it performs consistently well in several different real problems. Code for the presented algorithm is available at
Unified Structured Learning for Simultaneous Human Pose Estimation and Garment Attribute Classification
"... Abstract—In this paper, we utilize structured learning to simultaneously address two intertwined problems: human pose estimation (HPE) and garment attribute classification (GAC), which are valuable for a variety of computer vision and multimedia applications. Unlike previous works that usually han ..."
Abstract

Cited by 2 (0 self)
 Add to MetaCart
Abstract—In this paper, we utilize structured learning to simultaneously address two intertwined problems: human pose estimation (HPE) and garment attribute classification (GAC), which are valuable for a variety of computer vision and multimedia applications. Unlike previous works that usually handle the two problems separately, our approach aims to produce the optimal joint estimation for both HPE and GAC via a unified inference procedure. To this end, we adopt a preprocessing step to detect potential human parts from each image (i.e. a set of “candidates”) that allows us to have a manageable input space. In this way, the simultaneous inference of HPE and GAC is converted to a structured learning problem, where the inputs are the collections of candidate ensembles, the outputs are the joint labels of human parts and garment attributes, and the joint feature representation involves various cues such as posespecific features, garmentspecific features, and crosstask features that encode correlations between human parts and garment attributes. Furthermore, we explore the “strong edge ” evidence around the potential human parts so as to derive a more powerful representation for our oriented human part. The evidence can be seamlessly integrated into our structured learning model as a kind of energy function. The learning process is performed by the structured Support Vector Machines (SVM) algorithm. As the joint structure of the two problems is a cyclic graph, which hinders an efficient inference, we instead compute the approximate optima via an iterative process, where in each iteration the variables of one problem are fixed, i.e. an inference problem on a tree. In this way, the optimal solutions can be efficiently computed by a dynamic programming algorithm. Experimental results demonstrated on two benchmark datasets show the stateoftheart performance of our approach.
Learning to Search in BranchandBound Algorithms∗
"... Branchandbound is a widely used method in combinatorial optimization, including mixed integer programming, structured prediction and MAP inference. While most work has been focused on developing problemspecific techniques, little is known about how to systematically design the node searching str ..."
Abstract
 Add to MetaCart
(Show Context)
Branchandbound is a widely used method in combinatorial optimization, including mixed integer programming, structured prediction and MAP inference. While most work has been focused on developing problemspecific techniques, little is known about how to systematically design the node searching strategy on a branchandbound tree. We address the key challenge of learning an adaptive node searching order for any class of problem solvable by branchandbound. Our strategies are learned by imitation learning. We apply our algorithm to linear programming based branchandbound for solving mixed integer programs (MIP). We compare our method with one of the fastest opensource solvers, SCIP; and a very efficient commercial solver, Gurobi. We demonstrate that our approach achieves better solutions faster on four MIP libraries. 1
Tree Edges
"... GOAL: Propose an ecient and exact inference algorithm based on branchandbound (BB) to solve the human pose estimation problem on loopy graphical models Motivation: Cast human pose estimation problem as MAPMRFs inference problem Solving MAP inference on general MRFs is challenging Pros: ecient ..."
Abstract
 Add to MetaCart
(Show Context)
GOAL: Propose an ecient and exact inference algorithm based on branchandbound (BB) to solve the human pose estimation problem on loopy graphical models Motivation: Cast human pose estimation problem as MAPMRFs inference problem Solving MAP inference on general MRFs is challenging Pros: ecient inference by dynamic programming ( ). Cons: approximated model; common misclassication errors. Pros: more interactions between parts. Cons: exact inference NPhard; require approximated inference [1] or reduced state space. Contributions: Solve loopy models with large # of part hypotheses exactly. Eciency is provably guaranteed when # of nodes is small. Up to 74 times faster than competing techniques [2]! Enable more accurate results than nonloopy models [3]! Key Intuitions: Flexible bound by relaxing the loopy model into a mixture of star models. Novel data structure (BMT) and an ecient search routine (OBMS) to signicantly reduce the time complexity for calculating the bound in each branch of the BB search Given the model ,∑∑ εij ji p ij
DEDICATION
, 2012
"... This doctoral dissertation is dedicated to my ever supportive parents, loving wife and daughter. ii ACKNOWLEDGEMENTS I am very fortunate to work closely with both Professor Silvio Savarese and Professor FeiFei Li. They are both brilliant scientists and incredibly knowledgeable mentors. Through the ..."
Abstract
 Add to MetaCart
This doctoral dissertation is dedicated to my ever supportive parents, loving wife and daughter. ii ACKNOWLEDGEMENTS I am very fortunate to work closely with both Professor Silvio Savarese and Professor FeiFei Li. They are both brilliant scientists and incredibly knowledgeable mentors. Through the course of my PhD, they have helped me evolve from a clueless student to a confident researcher. I truly appreciate their guidances and encouragements. They have also taught my a lot on how to effectively communicate with other researchers: polishing research papers, and making presentation crystal clear and intuitive to follow. Most importantly, I kept on being amazed by their insights on highimpact future research directions, even after 5 years of study with them. I am also very lucky to study in both Princeton university and university of Michigan. At Princeton university, I have met many great professors. I want to thank them for teaching me the fundamental knowledge of machine learning and computer