The Experts below are selected from a list of 312 Experts worldwide ranked by ideXlab platform
Laurent Najman - One of the best experts on this subject based on the ideXlab platform.
-
Hierarchizing graph-based image segmentation algorithms relying on region dissimilarity: the case of the Felzenszwalb-Huttenlocher method
Mathematical Morphology - Theory and Applications, 2019Co-Authors: Silvio Jamil Ferzoli Guimarães, Yukiko Kenmochi, Jean Cousty, Zenilton Kleber Gonçalves Do Patrocínio, Laurent NajmanAbstract:The goal of this paper is to present an algorithm that builds a hierarchy of image seg-mentations from a class of dissimilarity criterions, the main example being the criterion proposed by Felzenszwalb and Huttenlocher which provides an observation scale. Specifically, we propose to select, for each observation scale, the largest not-too-coarse segmentation available in the hierarchy of quasi-flat zones. The resulting hierarchy is experimentally proved to be on par with the segmentation algorithm of Felzenszwalb and Huttenlocher, with the added property that it is now much easier to choose (tune) the scale of observation.
-
Causal graph-based video segmentation
2013Co-Authors: Camille Couprie, Clément Farabet, Yann Lecun, Laurent NajmanAbstract:Among the different methods producing superpixel segmentations of an image, the graph-based approach of Felzen-szwalb and Huttenlocher is broadly employed. One of its interesting properties is that the regions are computed in a greedy manner in quasi-linear time by using a minimum spanning tree. The algorithm may be trivially extended to video segmentation by considering a video as a 3D volume, however, this can not be the case for causal segmentation, when subsequent frames are unknown. In a framework exploiting minimum spanning trees all along, we propose an efficient video segmentation approach that computes temporally consistent pixels in a causal manner, filling the need for causal and real time applications.
-
ICIP - Causal graph-based video segmentation
2013 IEEE International Conference on Image Processing, 2013Co-Authors: Camille Couprie, Clément Farabet, Yann Lecun, Laurent NajmanAbstract:Among the different methods producing superpixel segmentations of an image, the graph-based approach of Felzenszwalb and Huttenlocher is broadly employed. One of its interesting properties is that the regions are computed in a greedy manner in quasi-linear time by using a minimum spanning tree. The algorithm may be trivially extended to video segmentation by considering a video as a 3D volume, however, this can not be the case for causal segmentation, when subsequent frames are unknown. In a framework exploiting minimum spanning trees all along, we propose an efficient video segmentation approach that computes temporally consistent pixels in a causal manner, filling the need for causal and real time applications.
-
A hierarchical image segmentation algorithm based on an observation scale
2012Co-Authors: Silvio Jamil Ferzoli Guimarães, Yukiko Kenmochi, Jean Cousty, Laurent NajmanAbstract:Hierarchical image segmentation provides a region-oriented scale-space, i.e., a set of image segmentations at different detail levels in which the segmentations at finer levels are nested with respect to those at coarser levels. Most image segmentation algorithms, such as region merging algorithms, rely on a criterion for merging that does not lead to a hierarchy. In addition, for image segmentation, the tuning of the parameters can be difficult. In this work, we propose a hierarchical graph based image segmentation relying on a criterion popularized by Felzenszwalb and Huttenlocher. Quantitative and qualitative assessments of the method on Berkeley image database shows efficiency, ease of use and robustness of our method.
-
AN EFFICIENT HIERARCHICAL GRAPH BASED IMAGE SEGMENTATION
arXiv: Computer Vision and Pattern Recognition, 2012Co-Authors: Silvio Jamil Ferzoli Guimarães, Yukiko Kenmochi, Jean Cousty, Laurent NajmanAbstract:Hierarchical image segmentation provides region-oriented scalespace, i.e., a set of image segmentations at different detail levels in which the segmentations at finer levels are nested with respect to those at coarser levels. Most image segmentation algorithms, such as region merging algorithms, rely on a criterion for merging that does not lead to a hierarchy, and for which the tuning of the parameters can be difficult. In this work, we propose a hierarchical graph based image segmentation relying on a criterion popularized by Felzenzwalb and Huttenlocher. We illustrate with both real and synthetic images, showing efficiency, ease of use, and robustness of our method. IndexTerms— Hierarchical image segmentation, Edge-weighted graph, Saliency map
Thomas F Shipley - One of the best experts on this subject based on the ideXlab platform.
-
application of the category adjustment model in temporal spatial and abstract magnitude at the billions scale
Cognitive Science, 2013Co-Authors: Ilyse Resnick, Thomas F ShipleyAbstract:Application of the Category Adjustment Model in Temporal, Spatial, and Abstract Magnitude at the Billions Scale Ilyse Resnick (ilyse.resnick@temple.edu) Department of Psychology, 1701 N. 13 th Street Philadelphia, PA 19122 USA Thomas F. Shipley (tshipley@temple.edu) Department of Psychology, 1701 N. 13 th Street Philadelphia, PA 19122 USA Abstract The current study examines the generalization of the Category Adjustment Model (CAM) across scales along two dimensions: time and distance. Participants were presented with geologic time and astronomical distance information either conventionally or using the hierarchical alignment model. Participants provided with hierarchically structured magnitude information for time and distances were more accurate on similar estimations at large scales than participants given the same content in a conventional manner. Patterns in event and distance estimation, along with overall group differences, are consistent with the CAM; suggesting people use hierarchically organized categorical information when estimating across scales and dimensions, and providing salient category boundary information improves estimation. Findings suggest a common representation of scale information for temporal, spatial, and abstract (numeric) magnitudes. Patterns of abstract magnitude estimations are consistent with segmented linear models of scale representation. Implications of the CAM in scale representation and the hierarchical alignment model in education are discussed. Keywords: Category Adjustment Alignment; Scale Representation Model; Hierarchical Introduction The Category Adjustment Model (CAM) is an adaptive Bayesian account for the pattern of systematic biases observed in recall of metric quantities due to category membership (Huttenlocher, Hedges, & Prohaska, 1988; Huttenlocher, Hedges, Vevea, 2000). The CAM posits 1D, 2D, and 3D magnitudes are stored in a hierarchical combination of metric and categorical information. In the absence of lower-level information (e.g., precise metric information), people use higher-level categories to aide in estimation. Variation in estimation, therefore, occurs due to imprecision of category boundaries. Recall is biased towards the ‘prototype’ of the respective category. For example, when recalling the position of an object in a circular display, participants naturally divide the circle into mental quadrants and the recalled location is biased towards the center (or prototype) of the relevant quadrant (Huttenlocher, Hedges, & Duncan, 1991). The CAM predicts recall patterns on a range of dimensions (e.g., fatness of fish, grayness of squares, and lengths of lines (Huttenlocher, et al., 2000), events (Huttenlocher, et al., 1988), and even social dimensions such as perception of facial expressions (Roberson, Damjanovic, & Pilling, 2007) and judgments of gender and ethnicity (Huart, Corneille, & Becquart, 2005)). However, there is limited research examining the CAM’s predictive capability for a given dimension (such as temporal and spatial scales) across different scales (such as from human scales through to scales outside of human perception). Science education research has identified conceptual categories for spatial and temporal scales outside of human perception (e.g., Trend, 2001; Tretter, Jones, Andre, Negishi, & Minogue, 2006), suggesting people may conceptualize magnitude information at relatively small and large temporal and spatial scales using a combination of metric and categorical information. Resnick, et al. (2012) experimentally assessed the role of categories in estimations of large temporal magnitudes. Participants who were provided with salient hierarchically organized event boundaries fostered a linear representation of events on the Geologic Time Scale compared to those who received the same information about the events without the salient hierarchical structure. Aligned with the CAM, this finding suggests the use of hierarchically organized category boundaries in the representation of events at larger temporal scales. The current study aims to add to this relatively sparse literature by examining the generalization of the CAM across scales and dimensions. Two main objectives are to replicate research on memory for large temporal magnitudes (geologic time), and extend research to another dimension: space. Astronomical distance (a spatial magnitude at a large scale) was chosen for two reasons. There is already extensive research on CAM and spatial distance; demonstrating spatial distances at familiar scales are stored in a combination of metric and categorical information (e.g., Huttenlocher, et al., 1991; Huttenlocher, et al., 2000). Additionally, while the precise nature of the relationship is unclear, there is a systematic relationship between time and distance (e.g., Clark, 1973; Gentner, 2001), suggesting that time and distance at human scales are represented and estimated in the same way. Thus, if temporal and spatial dimensions across familiar and relatively larger scales are represented in a similar way, an analogous pattern of memory performance would be expected.
-
CogSci - Application of the category adjustment model in temporal, spatial, and abstract magnitude at the billions scale
Cognitive Science, 2013Co-Authors: Ilyse Resnick, Thomas F ShipleyAbstract:Application of the Category Adjustment Model in Temporal, Spatial, and Abstract Magnitude at the Billions Scale Ilyse Resnick (ilyse.resnick@temple.edu) Department of Psychology, 1701 N. 13 th Street Philadelphia, PA 19122 USA Thomas F. Shipley (tshipley@temple.edu) Department of Psychology, 1701 N. 13 th Street Philadelphia, PA 19122 USA Abstract The current study examines the generalization of the Category Adjustment Model (CAM) across scales along two dimensions: time and distance. Participants were presented with geologic time and astronomical distance information either conventionally or using the hierarchical alignment model. Participants provided with hierarchically structured magnitude information for time and distances were more accurate on similar estimations at large scales than participants given the same content in a conventional manner. Patterns in event and distance estimation, along with overall group differences, are consistent with the CAM; suggesting people use hierarchically organized categorical information when estimating across scales and dimensions, and providing salient category boundary information improves estimation. Findings suggest a common representation of scale information for temporal, spatial, and abstract (numeric) magnitudes. Patterns of abstract magnitude estimations are consistent with segmented linear models of scale representation. Implications of the CAM in scale representation and the hierarchical alignment model in education are discussed. Keywords: Category Adjustment Alignment; Scale Representation Model; Hierarchical Introduction The Category Adjustment Model (CAM) is an adaptive Bayesian account for the pattern of systematic biases observed in recall of metric quantities due to category membership (Huttenlocher, Hedges, & Prohaska, 1988; Huttenlocher, Hedges, Vevea, 2000). The CAM posits 1D, 2D, and 3D magnitudes are stored in a hierarchical combination of metric and categorical information. In the absence of lower-level information (e.g., precise metric information), people use higher-level categories to aide in estimation. Variation in estimation, therefore, occurs due to imprecision of category boundaries. Recall is biased towards the ‘prototype’ of the respective category. For example, when recalling the position of an object in a circular display, participants naturally divide the circle into mental quadrants and the recalled location is biased towards the center (or prototype) of the relevant quadrant (Huttenlocher, Hedges, & Duncan, 1991). The CAM predicts recall patterns on a range of dimensions (e.g., fatness of fish, grayness of squares, and lengths of lines (Huttenlocher, et al., 2000), events (Huttenlocher, et al., 1988), and even social dimensions such as perception of facial expressions (Roberson, Damjanovic, & Pilling, 2007) and judgments of gender and ethnicity (Huart, Corneille, & Becquart, 2005)). However, there is limited research examining the CAM’s predictive capability for a given dimension (such as temporal and spatial scales) across different scales (such as from human scales through to scales outside of human perception). Science education research has identified conceptual categories for spatial and temporal scales outside of human perception (e.g., Trend, 2001; Tretter, Jones, Andre, Negishi, & Minogue, 2006), suggesting people may conceptualize magnitude information at relatively small and large temporal and spatial scales using a combination of metric and categorical information. Resnick, et al. (2012) experimentally assessed the role of categories in estimations of large temporal magnitudes. Participants who were provided with salient hierarchically organized event boundaries fostered a linear representation of events on the Geologic Time Scale compared to those who received the same information about the events without the salient hierarchical structure. Aligned with the CAM, this finding suggests the use of hierarchically organized category boundaries in the representation of events at larger temporal scales. The current study aims to add to this relatively sparse literature by examining the generalization of the CAM across scales and dimensions. Two main objectives are to replicate research on memory for large temporal magnitudes (geologic time), and extend research to another dimension: space. Astronomical distance (a spatial magnitude at a large scale) was chosen for two reasons. There is already extensive research on CAM and spatial distance; demonstrating spatial distances at familiar scales are stored in a combination of metric and categorical information (e.g., Huttenlocher, et al., 1991; Huttenlocher, et al., 2000). Additionally, while the precise nature of the relationship is unclear, there is a systematic relationship between time and distance (e.g., Clark, 1973; Gentner, 2001), suggesting that time and distance at human scales are represented and estimated in the same way. Thus, if temporal and spatial dimensions across familiar and relatively larger scales are represented in a similar way, an analogous pattern of memory performance would be expected.
Yeou-mei Christiana Liu - One of the best experts on this subject based on the ideXlab platform.
-
Medium-chain triglyceride (MCT) ketogenic therapy.
Epilepsia, 2008Co-Authors: Yeou-mei Christiana LiuAbstract:The medium-chain triglyceride diet (MCTD) is a variant of the classic 4:1 ketogenic diet (KD) introduced in 1971 by Huttenlocher as an attempt to improve the palatability of the KD by allowing more carbohydrates yet preserving ketosis. Although initially found to be equally effective as the classic KD, use of the MCTD declined because of frequent gastrointestinal side effects such as cramps, diarrhea, and vomiting. Recently, we have used the MCTD in more than 50 patients. We have found excellent seizure control, similar to the classic KD, and with careful monitoring, we have encountered minimal side effects. The MCTD should remain a viable dietary option for children with refractory epilepsy who have large appetites, can tolerate more calories, or cannot accept the restrictions of the classic KD.
-
Medium‐chain triglyceride (MCT) ketogenic therapy
Epilepsia, 2008Co-Authors: Yeou-mei Christiana LiuAbstract:The medium-chain triglyceride diet (MCTD) is a variant of the classic 4:1 ketogenic diet (KD) introduced in 1971 by Huttenlocher as an attempt to improve the palatability of the KD by allowing more carbohydrates yet preserving ketosis. Although initially found to be equally effective as the classic KD, use of the MCTD declined because of frequent gastrointestinal side effects such as cramps, diarrhea, and vomiting. Recently, we have used the MCTD in more than 50 patients. We have found excellent seizure control, similar to the classic KD, and with careful monitoring, we have encountered minimal side effects. The MCTD should remain a viable dietary option for children with refractory epilepsy who have large appetites, can tolerate more calories, or cannot accept the restrictions of the classic KD.
Silvio Jamil Ferzoli Guimarães - One of the best experts on this subject based on the ideXlab platform.
-
Hierarchizing graph-based image segmentation algorithms relying on region dissimilarity: the case of the Felzenszwalb-Huttenlocher method
Mathematical Morphology - Theory and Applications, 2019Co-Authors: Silvio Jamil Ferzoli Guimarães, Yukiko Kenmochi, Jean Cousty, Zenilton Kleber Gonçalves Do Patrocínio, Laurent NajmanAbstract:The goal of this paper is to present an algorithm that builds a hierarchy of image seg-mentations from a class of dissimilarity criterions, the main example being the criterion proposed by Felzenszwalb and Huttenlocher which provides an observation scale. Specifically, we propose to select, for each observation scale, the largest not-too-coarse segmentation available in the hierarchy of quasi-flat zones. The resulting hierarchy is experimentally proved to be on par with the segmentation algorithm of Felzenszwalb and Huttenlocher, with the added property that it is now much easier to choose (tune) the scale of observation.
-
Efficient Algorithms for Hierarchical Graph-Based Segmentation Relying on the Felzenszwalb–Huttenlocher Dissimilarity
International Journal of Pattern Recognition and Artificial Intelligence, 2019Co-Authors: Edward Jorge Yuri Cayllahua Cahuina, Yukiko Kenmochi, Jean Cousty, Arnaldo De Albuquerque Araujo, Guillermo Cámara-chávez, Silvio Jamil Ferzoli GuimarãesAbstract:Hierarchical image segmentation provides a region-oriented scale-space, i.e. a set of image segmentations at different detail levels in which the segmentations at finer levels are nested with respe...
-
efficient algorithms for hierarchical graph based segmentation relying on the felzenszwalb Huttenlocher dissimilarity
International Journal of Pattern Recognition and Artificial Intelligence, 2019Co-Authors: Edward Jorge Yuri Cayllahua Cahuina, Yukiko Kenmochi, Jean Cousty, Arnaldo De Albuquerque Araujo, Guillermo Camarachavez, Silvio Jamil Ferzoli GuimarãesAbstract:Hierarchical image segmentation provides a region-oriented scale-space, i.e. a set of image segmentations at different detail levels in which the segmentations at finer levels are nested with respe...
-
A hierarchical image segmentation algorithm based on an observation scale
2012Co-Authors: Silvio Jamil Ferzoli Guimarães, Yukiko Kenmochi, Jean Cousty, Laurent NajmanAbstract:Hierarchical image segmentation provides a region-oriented scale-space, i.e., a set of image segmentations at different detail levels in which the segmentations at finer levels are nested with respect to those at coarser levels. Most image segmentation algorithms, such as region merging algorithms, rely on a criterion for merging that does not lead to a hierarchy. In addition, for image segmentation, the tuning of the parameters can be difficult. In this work, we propose a hierarchical graph based image segmentation relying on a criterion popularized by Felzenszwalb and Huttenlocher. Quantitative and qualitative assessments of the method on Berkeley image database shows efficiency, ease of use and robustness of our method.
-
AN EFFICIENT HIERARCHICAL GRAPH BASED IMAGE SEGMENTATION
arXiv: Computer Vision and Pattern Recognition, 2012Co-Authors: Silvio Jamil Ferzoli Guimarães, Yukiko Kenmochi, Jean Cousty, Laurent NajmanAbstract:Hierarchical image segmentation provides region-oriented scalespace, i.e., a set of image segmentations at different detail levels in which the segmentations at finer levels are nested with respect to those at coarser levels. Most image segmentation algorithms, such as region merging algorithms, rely on a criterion for merging that does not lead to a hierarchy, and for which the tuning of the parameters can be difficult. In this work, we propose a hierarchical graph based image segmentation relying on a criterion popularized by Felzenzwalb and Huttenlocher. We illustrate with both real and synthetic images, showing efficiency, ease of use, and robustness of our method. IndexTerms— Hierarchical image segmentation, Edge-weighted graph, Saliency map
Jean Cousty - One of the best experts on this subject based on the ideXlab platform.
-
Hierarchizing graph-based image segmentation algorithms relying on region dissimilarity: the case of the Felzenszwalb-Huttenlocher method
Mathematical Morphology - Theory and Applications, 2019Co-Authors: Silvio Jamil Ferzoli Guimarães, Yukiko Kenmochi, Jean Cousty, Zenilton Kleber Gonçalves Do Patrocínio, Laurent NajmanAbstract:The goal of this paper is to present an algorithm that builds a hierarchy of image seg-mentations from a class of dissimilarity criterions, the main example being the criterion proposed by Felzenszwalb and Huttenlocher which provides an observation scale. Specifically, we propose to select, for each observation scale, the largest not-too-coarse segmentation available in the hierarchy of quasi-flat zones. The resulting hierarchy is experimentally proved to be on par with the segmentation algorithm of Felzenszwalb and Huttenlocher, with the added property that it is now much easier to choose (tune) the scale of observation.
-
Efficient Algorithms for Hierarchical Graph-Based Segmentation Relying on the Felzenszwalb–Huttenlocher Dissimilarity
International Journal of Pattern Recognition and Artificial Intelligence, 2019Co-Authors: Edward Jorge Yuri Cayllahua Cahuina, Yukiko Kenmochi, Jean Cousty, Arnaldo De Albuquerque Araujo, Guillermo Cámara-chávez, Silvio Jamil Ferzoli GuimarãesAbstract:Hierarchical image segmentation provides a region-oriented scale-space, i.e. a set of image segmentations at different detail levels in which the segmentations at finer levels are nested with respe...
-
efficient algorithms for hierarchical graph based segmentation relying on the felzenszwalb Huttenlocher dissimilarity
International Journal of Pattern Recognition and Artificial Intelligence, 2019Co-Authors: Edward Jorge Yuri Cayllahua Cahuina, Yukiko Kenmochi, Jean Cousty, Arnaldo De Albuquerque Araujo, Guillermo Camarachavez, Silvio Jamil Ferzoli GuimarãesAbstract:Hierarchical image segmentation provides a region-oriented scale-space, i.e. a set of image segmentations at different detail levels in which the segmentations at finer levels are nested with respe...
-
Efficient algorithms for hierarchical graph-based segmentation relying on the Felzenszwalb-Huttenlocher dissimilarity
International Journal of Pattern Recognition and Artificial Intelligence (IJPRAI), 2019Co-Authors: Edward Jorge Yuri Cayllahua Cahuina, Yukiko Kenmochi, Jean Cousty, Arnaldo De Albuquerque Araujo, Guillermo Cámara-chávez, Silvio Jamil F. GuimarãesAbstract:Hierarchical image segmentation provides a region-oriented scale-space, {\em i.e.}, a set of image segmentations at different detail levels in which the segmentations at finer levels are nested with respect to those at coarser levels. However, most image segmentation algorithms, among which a graph-based image segmentation method relying on a region merging criterion was proposed by Felzenszwalb-Huttenlocher in 2004, do not lead to a hierarchy. In order to cope with a demand for hierarchical segmentation, Guimar\~aes {\em et al.} proposed in 2012 a method for hierarchizing the popular Felzenszwalb-Huttenlocher method, without providing an algorithm to compute the proposed hierarchy. This article is devoted to provide a series of algorithms to compute the result of this hierarchical graph-based image segmentation method efficiently, based mainly on two ideas: optimal dissimilarity measuring and incremental update of the hierarchical structure. Experiments show that, for an image of size 321 $\times$ 481 pixels, the most efficient algorithm produces the result in half a second whereas the most naive one requires more than four hours.
-
Algorithms for hierarchical segmentation based on the Felzenszwalb-Huttenlocher dissimilarity
2018Co-Authors: Edward Jorge Yuri Cayllahua Cahuina, Yukiko Kenmochi, Jean Cousty, Arnaldo De Albuquerque Araujo, Guillermo Cámara-chávezAbstract:Hierarchical image segmentation provides a region-oriented scale-space, i.e., a set of image segmentations at different detail levels in which the segmentations at finer levels are nested with respect to those at coarser levels. Most image segmentation algorithms, such as region merging algorithms, rely on a criterion for merging that does not lead to a hierarchy. Guimarães et al. proposed in 2012 a hierarchical graph-based image segmentation method relying on a criterion popularized by Felzenszwalb and Huttenlocher in 2004, hence hierarchizing the popular Felzenszwalb-Huttenlocher method. However, Guimarães et al. did not provide an algorithm to compute the proposed hierarchy. We propose a series of algorithms to compute the result of this hierarchical graph-based image segmentation method. For an image of size 321 × 481 pixels, the most efficient algorithm produces the result in half a second whereas the most naive one requires more than four hours