The Experts below are selected from a list of 12672 Experts worldwide ranked by ideXlab platform
Freek Van Walderveen - One of the best experts on this subject based on the ideXlab platform.
-
locality and Bounding Box quality of two dimensional space filling curves
European Symposium on Algorithms, 2008Co-Authors: Herman Haverkort, Freek Van WalderveenAbstract:Space-filling curves can be used to organise points in the plane into Bounding-Box hierarchies (such as R-trees). We develop measures of the Bounding-Box qualityof space-filling curves that express how effective different curves are for this purpose. We give general lower bounds on the Bounding-Box quality and on locality according to Gotsman and Lindenbaum for a large class of curves. We describe a generic algorithm to approximate these and similar quality measures for any given curve. Using our algorithm we find good approximations of the locality and Bounding-Box quality of several known and new space-filling curves. Surprisingly, some curves with bad locality by Gotsman and Lindenbaum's measure, have good Bounding-Box quality, while the curve with the best-known locality has relatively bad Bounding-Box quality.
-
locality and Bounding Box quality of two dimensional space filling curves
arXiv: Computational Geometry, 2008Co-Authors: Herman Haverkort, Freek Van WalderveenAbstract:Space-filling curves can be used to organise points in the plane into Bounding-Box hierarchies (such as R-trees). We develop measures of the Bounding-Box quality of space-filling curves that express how effective different space-filling curves are for this purpose. We give general lower bounds on the Bounding-Box quality measures and on locality according to Gotsman and Lindenbaum for a large class of space-filling curves. We describe a generic algorithm to approximate these and similar quality measures for any given curve. Using our algorithm we find good approximations of the locality and the Bounding-Box quality of several known and new space-filling curves. Surprisingly, some curves with relatively bad locality by Gotsman and Lindenbaum's measure, have good Bounding-Box quality, while the curve with the best-known locality has relatively bad Bounding-Box quality.
Bernhard Kainz - One of the best experts on this subject based on the ideXlab platform.
-
deepcut object segmentation from Bounding Box annotations using convolutional neural networks
IEEE Transactions on Medical Imaging, 2017Co-Authors: Martin Rajchl, Matthew C.h. Lee, Ozan Oktay, Wenjia Bai, Konstantinos Kamnitsas, Mary A. Rutherford, Mellisa Damodaram, Joseph V Hajnal, Jonathan Passeratpalmbach, Bernhard KainzAbstract:In this paper, we propose DeepCut, a method to obtain pixelwise object segmentations given an image dataset labelled weak annotations, in our case Bounding Boxes. It extends the approach of the well-known GrabCut[1] method to include machine learning by training a neural network classifier from Bounding Box annotations. We formulate the problem as an energy minimisation problem over a densely-connected conditional random field and iteratively update the training targets to obtain pixelwise object segmentations. Additionally, we propose variants of the DeepCut method and compare those to a naive approach to CNN training under weak supervision. We test its applicability to solve brain and lung segmentation problems on a challenging fetal magnetic resonance dataset and obtain encouraging results in terms of accuracy.
-
DeepCut: Object Segmentation from Bounding Box Annotations Using Convolutional Neural Networks
IEEE Transactions on Medical Imaging, 2017Co-Authors: Martin Rajchl, Matthew C.h. Lee, Ozan Oktay, Wenjia Bai, Konstantinos Kamnitsas, Mary A. Rutherford, Mellisa Damodaram, Joseph V Hajnal, Jonathan Passerat-palmbach, Bernhard KainzAbstract:In this paper, we propose DeepCut, a method to obtain pixelwise object segmentations given an image dataset labelled with Bounding Box annotations. It extends the approach of the well-known GrabCut method to include machine learning by training a neural network classifier from Bounding Box annotations. We formulate the problem as an energy minimisation problem over a densely-connected conditional random field and iteratively update the training targets to obtain pixelwise object segmentations. Additionally, we propose variants of the DeepCut method and compare those to a naive approach to CNN training under weak supervision. We test its applicability to solve brain and lung segmentation problems on a challenging fetal magnetic resonance dataset and obtain encouraging results in terms of accuracy.
Stefan Vandewalle - One of the best experts on this subject based on the ideXlab platform.
-
Bounding Box framework for efficient phase field simulation of grain growth in anisotropic systems
Computational Materials Science, 2011Co-Authors: Liesbeth Vanherpe, Nele Moelans, Bart Blanpain, Stefan VandewalleAbstract:A sparse Bounding Box algorithm is extended to perform efficient phase field simulations of grain growth in anisotropic systems. The extended Bounding Box framework allows to attribute different properties to different grain boundary types of a polycrystalline microstructure and can be combined with explicit, implicit or semi-implicit time stepping strategies. To illustrate the applicability of the software, the simulation results of a case study are analysed. They indicate the impact of a misorientation dependent boundary energy formulation on the evolution of the misorientation distribution of the grain boundary types and on the individual growth rates of the grains as a function of the number of grain faces.
-
Bounding Box algorithm for three dimensional phase field simulations of microstructural evolution in polycrystalline materials
Physical Review E, 2007Co-Authors: Liesbeth Vanherpe, Nele Moelans, Bart Blanpain, Stefan VandewalleAbstract:Phase-field modeling has proven to be a versatile tool for simulating microstructural evolution phenomena, such as grain growth in polycrystalline materials. However, the computing time and computing memory requirements of a phase-field model pose severe limitations on the number of phase-field variables that can be taken into account in a practical implementation. In this paper, a sparse Bounding Box algorithm is proposed that allows the use of a large number of phase-field variables without excessive memory usage or computational requirements. The algorithm is applied to a three-dimensional model for grain growth in the presence of second-phase particles.
Hua Zong - One of the best experts on this subject based on the ideXlab platform.
-
a novel cnn based method for accurate ship detection in hr optical remote sensing images via rotated Bounding Box
IEEE Transactions on Geoscience and Remote Sensing, 2021Co-Authors: Zhiqiang Zhou, Bo Wang, Lingjuan Miao, Hua ZongAbstract:Currently, reliable and accurate ship detection in optical remote sensing images is still challenging. Even the state-of-the-art convolutional neural network (CNN)-based methods cannot obtain very satisfactory results. To more accurately locate the ships in diverse orientations, some recent methods conduct the detection via the rotated Bounding Box. However, it further increases the difficulty of detection because an additional variable of ship orientation must be accurately predicted in the algorithm. In this article, a novel CNN-based ship-detection method is proposed by overcoming some common deficiencies of current CNN-based methods in ship detection. Specifically, to generate rotated region proposals, current methods have to predefine multioriented anchors and predict all unknown variables together in one regression process, limiting the quality of overall prediction. By contrast, we are able to predict the orientation and other variables independently, and yet more effectively, with a novel dual-branch regression network, based on the observation that the ship targets are nearly rotation-invariant in remote sensing images. Next, a shape-adaptive pooling method is proposed to overcome the limitation of a typical regular region of interest (ROI) pooling in extracting the features of the ships with various aspect ratios. Furthermore, we propose to incorporate multilevel features via the spatially variant adaptive pooling. This novel approach, called multilevel adaptive pooling, leads to a compact feature representation more qualified for the simultaneous ship classification and localization. Finally, a detailed ablation study performed on the proposed approaches is provided, along with some useful insights. Experimental results demonstrate the great superiority of the proposed method in ship detection.
-
a novel cnn based method for accurate ship detection in hr optical remote sensing images via rotated Bounding Box
arXiv: Computer Vision and Pattern Recognition, 2020Co-Authors: Zhiqiang Zhou, Bo Wang, Lingjuan Miao, Hua ZongAbstract:Currently, reliable and accurate ship detection in optical remote sensing images is still challenging. Even the state-of-the-art convolutional neural network (CNN) based methods cannot obtain very satisfactory results. To more accurately locate the ships in diverse orientations, some recent methods conduct the detection via the rotated Bounding Box. However, it further increases the difficulty of detection, because an additional variable of ship orientation must be accurately predicted in the algorithm. In this paper, a novel CNN-based ship detection method is proposed, by overcoming some common deficiencies of current CNN-based methods in ship detection. Specifically, to generate rotated region proposals, current methods have to predefine multi-oriented anchors, and predict all unknown variables together in one regression process, limiting the quality of overall prediction. By contrast, we are able to predict the orientation and other variables independently, and yet more effectively, with a novel dual-branch regression network, based on the observation that the ship targets are nearly rotation-invariant in remote sensing images. Next, a shape-adaptive pooling method is proposed, to overcome the limitation of typical regular ROI-pooling in extracting the features of the ships with various aspect ratios. Furthermore, we propose to incorporate multilevel features via the spatially-variant adaptive pooling. This novel approach, called multilevel adaptive pooling, leads to a compact feature representation more qualified for the simultaneous ship classification and localization. Finally, detailed ablation study performed on the proposed approaches is provided, along with some useful insights. Experimental results demonstrate the great superiority of the proposed method in ship detection.
Herman Haverkort - One of the best experts on this subject based on the ideXlab platform.
-
locality and Bounding Box quality of two dimensional space filling curves
European Symposium on Algorithms, 2008Co-Authors: Herman Haverkort, Freek Van WalderveenAbstract:Space-filling curves can be used to organise points in the plane into Bounding-Box hierarchies (such as R-trees). We develop measures of the Bounding-Box qualityof space-filling curves that express how effective different curves are for this purpose. We give general lower bounds on the Bounding-Box quality and on locality according to Gotsman and Lindenbaum for a large class of curves. We describe a generic algorithm to approximate these and similar quality measures for any given curve. Using our algorithm we find good approximations of the locality and Bounding-Box quality of several known and new space-filling curves. Surprisingly, some curves with bad locality by Gotsman and Lindenbaum's measure, have good Bounding-Box quality, while the curve with the best-known locality has relatively bad Bounding-Box quality.
-
locality and Bounding Box quality of two dimensional space filling curves
arXiv: Computational Geometry, 2008Co-Authors: Herman Haverkort, Freek Van WalderveenAbstract:Space-filling curves can be used to organise points in the plane into Bounding-Box hierarchies (such as R-trees). We develop measures of the Bounding-Box quality of space-filling curves that express how effective different space-filling curves are for this purpose. We give general lower bounds on the Bounding-Box quality measures and on locality according to Gotsman and Lindenbaum for a large class of space-filling curves. We describe a generic algorithm to approximate these and similar quality measures for any given curve. Using our algorithm we find good approximations of the locality and the Bounding-Box quality of several known and new space-filling curves. Surprisingly, some curves with relatively bad locality by Gotsman and Lindenbaum's measure, have good Bounding-Box quality, while the curve with the best-known locality has relatively bad Bounding-Box quality.