Combining Top-Down and Bottom-Up Segmentation

By Eran Borenstein, Eitan Sharon, Shimon Ullman. 2004

354 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Perceiver: General Perception with Iterative Attention (2021, background): “Re-entrant processing of this kind (sometimes referred to as top-down processing) has a long history in computer vision (Borenstein et al., 2004; Kumar et al., 2005; Carreira et al., 2016; Hu & Ramanan, 2016; Yang et al., 2018; Lin et al., 2020).”
Using Multiple Segmentations to Discover Objects and their Extent in Image Collections (2006, methodology): “This idea of maintaining multiple segmentations until further evidence can be used to disambiguate is similar to the approach of Borenstein et al. [ 3 ].”
DeepEdge: A multi-scale bifurcated deep network for top-down contour detection (2014, methodology): “Contour detection is typically considered a low-level problem, and used to aid higher-level tasks such as object detection [2, 3, 27, 21].”
ReSeg: A Recurrent Neural Network for Object Segmentation (2015, methodology): “We test the performances of the model in the object segmentation domain on one of the historically most used datasets in this field, the Weizmann Horse dataset (Borenstein, 2004), the Oxford Flowers 17 dataset (Nilsback & Zisserman, 2006) and the more recent Fashionista dataset (Yamaguchi, 2012).”
Auto-context and its application to high-level vision tasks (2008, methodology): “The first and the fifth column displays some test images from the Weizmann dataset [3].”
LOCUS: learning object classes with unsupervised segmentation (2005, methodology): “For comparison, the table also includes results from Borenstein et al [4] whose method used 54 hand-segmented training images.”
ImageNet Auto-Annotation with Segmentation Propagation (2014, background): “Object segmentation Fully supervised segmentation techniques aim at separating instances of an object class from their background (e.g. horses, faces, cars [11,33,9]).”
Groups of Adjacent Contour Segments for Object Detection (2008, methodology): “The positive images are derived from a data set previously released by the Weizmann Institute for evaluating image segmentation algorithms [ 2 ].”
Contour-based learning for object detection (2005, methodology): “Borenstein & Ullman [ 2 ] present good class-specific segmentation results but do not evaluate their technique in classification or detection performance, or indeed show results on images not containing objects of the trained class.”
CPMC: Automatic Object Segmentation Using Constrained Parametric Min-Cuts (2012, neutral): “Expressive part-based appearance models have also been developed [49], [50], [51], [52].”
Active skeleton for non-rigid object detection (2009, methodology): “We provide experimental evaluation of the proposed approach on two datasets: the Weizmann horses [7] and the ETHZ shapes [16].”
Robust Object Detection with Interleaved Categorization and Segmentation (2008, positive): “In more recent work, the approach is extended to also combine the top-down segmentation with bottom-up segmentation cues in order to obtain higher-quality results ( Borenstein et al. 2004 ).”
TextonBoost: Joint Appearance, Shape and Context Modeling for Multi-class Object Recognition and Segmentation (2006, neutral): “[5, 6].”
Semantic Image Segmentation with Contextual Hierarchical Models (2015, methodology): “In extensive experiments, we demonstrate the performance of CHM on a couple of challenging vision tasks: Horse segmentation in the Weizmann dataset [17], outdoor scene labeling in the Stanford background [18].”