Beyond Gradient Descent for Regularized Segmentation Losses

Marin, Dmitrii; Tang, Meng; Ayed, Ismail Ben; Boykov, Yuri

doi:10.1109/cvpr.2019.01043

Cited by 24 publications

(22 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To further improve the quality of weaklysupervised training, it is possible to leverage standard lowlevel regularizes over a large number of unlabeled pixels. [66,32,61,62,42]. For example, [62] achieves the stateof-the-art using bilinear relaxation of the Potts model in (3)…”

Section: Regularized Losses In Cnn Segmentationmentioning

confidence: 99%

Robust Trust Region for Weakly Supervised Segmentation

Marin¹,

Boykov²

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Acquisition of training data for the standard semantic segmentation is expensive if requiring that each pixel is labeled. Yet, current methods significantly deteriorate in weakly supervised settings, e.g. where a fraction of pixels is labeled or when only image-level tags are available. It has been shown that regularized losses-originally developed for unsupervised low-level segmentation and representing geometric priors on pixel labels-can considerably improve the quality of weakly supervised training. However, many common priors require optimization stronger than gradient descent. Thus, such regularizers have limited applicability in deep learning. We propose a new robust trust region approach for regularized losses improving the state-of-the-art results. Our approach can be seen as a higher-order generalization of the classic chain rule. It allows neural network optimization to use strong low-level solvers for the corresponding regularizers, including discrete ones.

show abstract

Section: Regularized Losses In Cnn Segmentationmentioning

confidence: 99%

Robust Trust Region for Weakly Supervised Segmentation

Marin¹,

Boykov²

2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Alternatively, one can directly use objective (2) as a regularized loss function [50,52]. Our proposal generation approach can be seen as a one step of ADM procedure for such a loss [37].…”

Section: Sampling Modelmentioning

confidence: 99%

Efficient Segmentation: Learning Downsampling Near Semantic Boundaries

Marin

Vajda

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

Self Cite

View full text Add to dashboard Cite

Many automated processes such as auto-piloting rely on a good semantic segmentation as a critical component. To speed up performance, it is common to downsample the input frame. However, this comes at the cost of missed small objects and reduced accuracy at semantic boundaries. To address this problem, we propose a new content-adaptive downsampling technique that learns to favor sampling locations near semantic boundaries of target classes. Costperformance analysis shows that our method consistently outperforms the uniform sampling improving balance between accuracy and computational efficiency. Our adaptive sampling gives segmentation with better quality of boundaries and more reliable support for smaller-size objects.

show abstract

“…Size-constrained proposal e y < l a t e x i t s h a 1 _ b a s e 6 4 = " C P 7 J n V i B w g n m j Y I 5 t 4 ADMM for incorporating high-order segmentation priors on the target region's histogram of intensities [26] or compactness [17]. More recently, similar techniques have been proposed to include higher-order priors directly in the learning process [27,28]. To our knowledge, our work is the first employing a discretecontinuous framework for weakly-supervised segmentation.…”

Section: Crf-regularized Proposalmentioning

confidence: 99%

Discretely-constrained deep network for weakly supervised segmentation

Peng

Kervadec

Dolz³

et al. 2019

Preprint

Self Cite

View full text Add to dashboard Cite

An efficient strategy for weakly-supervised segmentation is to impose constraints or regularization priors on target regions. Recent efforts have focused on incorporating such constraints in the training of convolutional neural networks (CNN), however this has so far been done within a continuous optimization framework. Yet, various segmentation constraints and regularization can be modeled and optimized more efficiently in a discrete formulation. This paper proposes a method, based on the alternating direction method of multipliers (ADMM) algorithm, to train a CNN with discrete constraints and regularization priors. This method is applied to the segmentation of medical images with weak annotations, where both size constraints and boundary length regularization are enforced. Experiments on two benchmark datasets for medical image segmentation show our method to provide significant improvements compared to existing approaches in terms of segmentation accuracy, constraint satisfaction and convergence speed.

show abstract

Beyond Gradient Descent for Regularized Segmentation Losses

Cited by 24 publications

References 47 publications

Robust Trust Region for Weakly Supervised Segmentation

Robust Trust Region for Weakly Supervised Segmentation

Efficient Segmentation: Learning Downsampling Near Semantic Boundaries

Discretely-constrained deep network for weakly supervised segmentation

Contact Info

Product

Resources

About