Learning optimized MAP estimates in continuously-valued MRF models

Samuel, Kegan G. G.; Tappen, Marshall F.

doi:10.1109/cvpr.2009.5206774

Cited by 67 publications

(72 citation statements)

References 16 publications

(15 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We incorporate this data-driven regularization model into a standard Markov Random Field (MRF) model, inferred with a gradient descent algorithm [14] and learned with a discriminative learning approach [19]. Compared to prior state-of-the-art methods, our full model achieves comparable or better results on the KITTI stereo and flow datasets [7], and improves results on the Sintel Flow dataset [4] under an online estimation setting.…”

Section: Introductionmentioning

confidence: 96%

“…We do a grid search for λ S and below we describe the discriminative learning approach [19] to optimize {γ q }.…”

Section: Learningmentioning

confidence: 99%

“…In this paper, we present a new regularization model for stereo or flow through transferring the shape information of the disparity or flow from semantically matched patches in the training database. Compared to previous regularization models based on image appearance alone, we can better resolve local ambiguity of the disparity or flow by considering the semantic information without explicit object modeling [1,5].We incorporate this data-driven regularization model into a standard Markov Random Field (MRF) model, inferred with a gradient descent algorithm [14] and learned with a discriminative learning approach [19]. Compared to prior state-of-the-art methods, our full model achieves comparable or better results on the KITTI stereo and flow datasets [7], and improves results on the Sintel Flow dataset [4] under an online estimation setting.…”

mentioning

confidence: 99%

See 2 more Smart Citations

A Data-Driven Regularization Model for Stereo and Flow

Wei

Liu

Freeman

2014

2014 2nd International Conference on 3D Vision

View full text Add to dashboard Cite

Data-driven techniques can reliably build semantic correspondence among images [16,15]. In this paper, we present a new regularization model for stereo or flow through transferring the shape information of the disparity or flow from semantically matched patches in the training database. Compared to previous regularization models based on image appearance alone, we can better resolve local ambiguity of the disparity or flow by considering the semantic information without explicit object modeling [1,5].We incorporate this data-driven regularization model into a standard Markov Random Field (MRF) model, inferred with a gradient descent algorithm [14] and learned with a discriminative learning approach [19]. Compared to prior state-of-the-art methods, our full model achieves comparable or better results on the KITTI stereo and flow datasets [7], and improves results on the Sintel Flow dataset [4] under an online estimation setting.

show abstract

Section: Introductionmentioning

confidence: 96%

“…We do a grid search for λ S and below we describe the discriminative learning approach [19] to optimize {γ q }.…”

Section: Learningmentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

A Data-Driven Regularization Model for Stereo and Flow

Wei

Liu

Freeman

2014

2014 2nd International Conference on 3D Vision

View full text Add to dashboard Cite

show abstract

“…In [32], Tappen trained FOE parameters by minimizing a loss function with stochastic gradient descent. Samuel and Tappen [23] presented an improved version based on implicit-differentiation. Li and Huttenlocher [18] presented a discriminative learning algorithm based on simultaneous perturbation stochastic approximation.…”

Section: Related Workmentioning

confidence: 99%

“…Since then however, a large body of work [21,34,36,38] has found that histograms of filter responses for natural images tend to be highly "nonGaussian", in that they have sharp peaks at zero and heavy tails. Consequently, recent works have focused on nonconvex priors [2,22,23,32,36].…”

Section: Introductionmentioning

confidence: 99%

Learning the right model: Efficient max-margin learning in Laplacian CRFs

Batra

Saxena

2012

2012 IEEE Conference on Computer Vision and Pattern Recognition

View full text Add to dashboard Cite

An important modeling decision made while designing Conditional Random Fields (CRFs) is the choice of the potential functions over the cliques of variables. Laplacian potentials are useful because they are robust potentials and match image statistics better than Gaussians. Moreover, energies with Laplacian terms remain convex, which simplifies inference. This makes Laplacian potentials an ideal modeling choice for some applications.In this paper, we study max-margin parameter learning in CRFs with Laplacian potentials (LCRFs). We first show that structured hinge-loss [35] is non-convex for LCRFs and thus techniques used by previous works are not applicable. We then present the first approximate max-margin algorithm for LCRFs. Finally, we make our learning algorithm scalable in the number of training images by using dual-decomposition techniques. Our experiments on singleimage depth estimation show that even with simple features, our approach achieves comparable to state-of-art results.

show abstract