Mind the Gap: Modeling Local and Global Context in (Road) Networks

Montoya-Zegarra, Javier A.; Wegner, Jan Dirk; Ladický, Ľubor; Schindler, Konrad

doi:10.1007/978-3-319-11752-2_17

Cited by 40 publications

(21 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Next, they extract either high-tubularity superpixels likely to be tubular structure fragments [SPHHP + 15,ZWLS14] or longer paths connecting points likely to be on the centerline of such structures [GFF08, BSBZ13, NGN + 15, TBA + 16]. Each superpixel or path is treated as an edge e i of an over-complete spatial graph G (see Fig.…”

Section: Graph-based Delineationmentioning

confidence: 99%

Active Learning and Proofreading for Delineation of Curvilinear Structures

Mosinska

Tarnawski

Fua

2017

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Many state-of-the-art delineation methods rely on supervised machine learning algorithms. As a result, they require manually annotated training data, which is tedious to obtain. Furthermore, even minor classification errors may significantly affect the topology of the final result. In this paper we propose a generic approach to addressing both of these problems by taking into account the influence of a potential misclassification on the resulting delineation. In an Active Learning context, we identify parts of linear structures that should be annotated first in order to train a classifier effectively. In a proofreading context, we similarly find regions of the resulting reconstruction that should be verified in priority to obtain a nearly-perfect result. In both cases, by focusing the attention of the human expert on potential classification mistakes which are the most critical parts of the delineation, we reduce the amount of required supervision. We demonstrate the effectiveness of our approach on microscopy images depicting blood vessels and neurons.

show abstract

Section: Graph-based Delineationmentioning

confidence: 99%

Active Learning and Proofreading for Delineation of Curvilinear Structures

Mosinska

Tarnawski

Fua

2017

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…They not only recover the geometry of the problem, but also the correct connectivity, which is crucial in applications such as neuroscience [21,29,18,28,27,20]. They largely owe their performance to supervised Machine Learning techniques that allow them to recognize promising linear paths.…”

Section: Active Learning For Delineationmentioning

confidence: 99%

“…Next, a set of subsampled high-tubularity superpixels [21,29,18] or longer paths [28,27,3,20] are extracted. Each of them can be considered as an edge e i belonging to overcomplete spatial graph G and characterized by a feature vector x i .…”

Section: Active Learning For Delineationmentioning

confidence: 99%

“…They range in size from solar filaments as seen through telescopes to road networks in aerial images, blood vessels in medical imagery, and neural structures in micrographs. These very diverse structures have different appearances and it has recently been shown that training classifiers to assess whether an image path is likely to be a structure of interest is key to improving the performance of automated delineation algorithms [28,27,3,21,18,29].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Active Learning for Delineation of Curvilinear Structures

Mosinska-Domanska¹,

Sznitman

Głowacki

et al. 2016

2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

View full text Add to dashboard Cite

Many recent delineation techniques owe much of their increased effectiveness to path classification algorithms that make it possible to distinguish promising paths from others. The downside of this development is that they require annotated training data, which is tedious to produce.In this paper, we propose an Active Learning approach that considerably speeds up the annotation process. Unlike standard ones, it takes advantage of the specificities of the delineation problem. It operates on a graph and can reduce the training set size by up to 80% without compromising the reconstruction quality.We will show that our approach outperforms conventional ones on various biomedical and natural image datasets, thus showing that it is broadly applicable.

show abstract

“…blood vessel detection from medical images, is also a well-known issue in computer vision (Chai et al, 2013). Chai et al (2013) distinguished three types of approach: pixel-based (Stoica et al, 2004;Mnih and Hinton, 2010;Montoya-Zegarra et al, 2014), line-based (Lacoste et al, 2005), and graph-based (Gerke et al, 2004;Turetken et al, 2012;Ünsalan and Sirmacek, 2012;Turetken et al, 2013;Wegner et al, 2013). A road extraction system often involves several different types of techniques.…”

Section: Introductionmentioning

confidence: 99%

Urban land use extraction from very high resolution remote sensing images

View full text Add to dashboard Cite

Mind the Gap: Modeling Local and Global Context in (Road) Networks

Cited by 40 publications

References 30 publications

Active Learning and Proofreading for Delineation of Curvilinear Structures

Active Learning and Proofreading for Delineation of Curvilinear Structures

Active Learning for Delineation of Curvilinear Structures

Urban land use extraction from very high resolution remote sensing images

Contact Info

Product

Resources

About