Efficient Optimization for Hierarchically-Structured Interacting Segments (HINTS)

@article{Isack2017EfficientOF,
  title={Efficient Optimization for Hierarchically-Structured Interacting Segments (HINTS)},
  author={Hossam N. Isack and Olga Veksler and Ipek Oguz and Milan Sonka and Yuri Boykov},
  journal={2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2017},
  pages={4981-4989}
}
We propose an effective optimization algorithm for a general hierarchical segmentation model with geometric interactions between segments. Any given tree can specify a partial order over object labels defining a hierarchy. It is well-established that segment interactions, such as inclusion/exclusion and margin constraints, make the model significantly more discriminant. However, existing optimization methods do not allow full use of such models. Generic a-expansion results in weak local minima… 
Sparse Layered Graphs for Multi-Object Segmentation
TLDR
This work forms a general method for adding containment and exclusion interaction constraints to layered graphs, and proposes three algorithms for further reducing the spatial complexity of SLGs, by using ordered multi-column graphs.
SOL: Segmentation with Overlapping Labels
  • K. Ng
  • Computer Science
  • 2018
TLDR
It is proved that SOL is an NP-hard problem, as well as a novel move-making optimization framework to find an approximate solution to SOL, and the proposed method outperforms state-of-the-art algorithms for multi-part segmentation.
Bayesian Polytrees With Learned Deep Features for Multi-Class Cell Segmentation
TLDR
Polytrees are shown to outperform directed trees in predicting segmentation error by highlighting areas in the segmented image that do not comply with prior knowledge, paving the way to uncertainty measures on the resulting segmentation and guide subsequent segmentation refinement.
Repository January 2018 SOL : Segmentation with Overlapping Labels
  • K. Ng
  • Computer Science
  • 2018
TLDR
It is proved that SOL is an NP-hard problem, as well as a novel move-making optimization framework to find an approximate solution to SOL, and the proposed method outperforms state-of-the-art algorithms for multi-part segmentation.
MRF Optimization with Separable Convex Prior on Partially Ordered Labels
TLDR
This paper introduces a generalization to partially ordered sets and proposes an efficient coarse-to-fine approach over the label space that provides an approximate solution to the multi-labeling problem with a convex penalty.
Double Expansion for Multilabel Segmentation with Shape Prior
TLDR
A new move making algorithm, called double expansion, is developed that allows each pixel to choose a label from a pair of new labels or keep the old label, resulting in an algorithm with optimality guarantees and robust performance in practice.
K-convexity Shape Priors for Segmentation
TLDR
For many forms of convexity the authors' regularization model is significantly more descriptive for any given k, and in practice this shape prior is useful in practice, e.g. biomedical applications, and its optimization is robust to local minima.
Region growing using superpixels with learned shape prior
TLDR
The performance of the proposed method is demonstrated and it is compared with alternative approaches on the task of segmenting individual eggs in microscopy images of Drosophila ovaries on the level of superpixels instead of pixels.
Faster Multi-Object Segmentation using Parallel Quadratic Pseudo-Boolean Optimization
TLDR
A parallel version of the Quadratic Pseudo-Boolean Optimization (QPBO) algorithm for solving binary optimization tasks, such as image segmentation, that is over 20 times faster than the serial QPBO algorithm on the large tasks and over three times faster for the majority of the small tasks.
Review of Serial and Parallel Min-Cut/Max-Flow Algorithms for Computer Vision
TLDR
Evaluating the state-of-the-art serial and parallel min-cut/max-flow algorithms on the largest set of computer vision problems yet finds the bottom-up merging approach by Liu and Sun to be best, but no method is dominant.
...
...

References

SHOWING 1-10 OF 30 REFERENCES
Graph cut with ordering constraints on labels and its applications
TLDR
It is observed that the commonly used graph-cut based alpha-expansion is more likely to get stuck in a local minimum when ordering constraints are used, so order-preserving moves are developed, which are developed and used for certain simple shape priors in graphcut segmentation.
Hedgehog Shape Priors for Multi-Object Segmentation
TLDR
This work proposes a more general multi-object segmentation approach that has significantly more applications than standard single star-convex segmentation, e.g. in medical data the authors can separate multiple non-star organs with similar appearances and weak edges, and modified -expansion moves shown to be submodular for multi-hedgehog shapes.
Optimizing Binary MRFs via Extended Roof Duality
TLDR
An efficient implementation of the "probing" technique is discussed, which simplifies the MRF while preserving the global optimum, and a new technique which takes an arbitrary input labeling and tries to improve its energy is presented.
Globally optimal pixel labeling algorithms for tree metrics
TLDR
This work substantially improves a facility location algorithm of Kolen, which is impractical for large label sets L since it requires O(|L|) min cuts on large graphs, and provides fast algorithms that use graph cuts to exactly minimize the energy function for pixel labeling problems with tree metrics.
"GrabCut": interactive foreground extraction using iterated graph cuts
TLDR
A more powerful, iterative version of the optimisation of the graph-cut approach is developed and the power of the iterative algorithm is used to simplify substantially the user interaction needed for a given quality of result.
Graph Cuts and Efficient N-D Image Segmentation
TLDR
This application epitomizes the best features of combinatorial graph cuts methods in vision: global optima, practical efficiency, numerical robustness, ability to fuse a wide range of visual cues and constraints, unrestricted topological properties of segments, and applicability to N-D problems.
Potts Model, Parametric Maxflow and K-Submodular Functions
TLDR
The technique to reduce the runtime to O(log k) maxflow computations (or one parametric maxflow computation) is shown, which allows to speed-up the subsequent alpha expansion for the unlabeled part, or can be used as it is for time-critical applications.
Fast Approximate Energy Minimization with Label Costs
TLDR
The main algorithmic contribution is an extension of α-expansion that also optimizes “label costs” with well-characterized optimality bounds, which has a natural interpretation as minimizing description length (MDL) and sheds light on classical algorithms like K-means and expectation-maximization (EM).
Energy-Based Geometric Multi-model Fitting
TLDR
The proposed PEaRL combines model sampling from data points as in RANSAC with iterative re-estimation of inliers and models’ parameters based on a global regularization functional and converges to a good quality local minimum of the energy automatically selecting a small number of models that best explain the whole data set.
An Efficient Optimization Framework for Multi-Region Segmentation Based on Lagrangian Duality
TLDR
A multi-region model for simultaneous segmentation of medical images with geometric constraints such as inclusion and exclusion between the regions are enforced, which makes it possible to correctly segment different regions even if the intensity distributions are identical.
...
...