Using contours to detect and localize junctions in natural images

By Michael Maire, Pablo Arbeláez, Charless C. Fowlkes, Jitendra Malik. 2008 IEEE Conference on Computer Vision and Pattern Recognition, 2008

476 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/4bb0f3a570936826401b4d1d322725bec3267dce

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

Unsupervised Template Learning for Fine-Grained Object Recognition (2012, methodology): “We first run the Berkeley edge detector [25] to compute the edge map of an image, and then treat it as a grayscale image and extract color kernel descriptors [3] over it.”
CPMC: Automatic Object Segmentation Using Constrained Parametric Min-Cuts (2012, methodology): “They use the output of the learned globalPb boundary detector [34] and can represent the full hierarchy elegantly by a single ultrametric contour map.”
Poselets: Body part detectors trained using 3D human pose annotations (2009, methodology): “Bottom: Our region labelling tool performs hierarchical oversegmentation of the image using [9] to allow the user to efficiently and accurately assign region labels.”
Depth Transfer: Depth Extraction from Video Using Non-Parametric Sampling (2014, methodology): “We determine this weighting by applying a sigmoidal function to the gradients, which we found to produce more pleasing inferred depth maps than using other boundary detecting schemes such as [32], [33].”
DeepContour: A deep convolutional feature learned by positive-sharing loss for contour detection (2015, background): “However, obvious performance gap is still observed between contour detections given by the algorithms and human annotations.”
From contours to regions: An empirical evaluation (2009, methodology): “The best detector, gPb [23], does not produce closed boundaries required to form a segmentation.”
Learning full pairwise affinities for spectral segmentation (2010, methodology): “2a by Canny-OWT-UCM, gPb-OWT-UCM [25], and our algorithm fPb-OWT-UCM, respectively.”
Large displacement optical flow (2009, methodology): “The segmentation is based on the boundary detector gPb from [11].”
Segmentation-Aware Convolutional Networks Using Local Attention Masks (2017, methodology): “For instance in [54, 55], soft segmentation masks [41, 32] and boundary cues [36, 48] have been used to develop segmentation-aware variants of handcrafted features, like SIFT and HOG, effectively suppressing contributions from pixels likely to come from the background.”
Continuous Markov Random Fields for Robust Stereo Estimation (2012, background): “The constraints on label combinations at junctions often force the labeling of the entire line drawing [14].”
Asymmetric Cuts: Joint Image Labeling and Partitioning (2014, background): “probability of boundary (gPb) detector from [29].”
A Particle Filter Framework for Contour Detection (2012, methodology): “To evaluate our algorithm, we replicate the scenario used in the evaluation of state-ofthe-art contour detection methods [2–9], using the BSDS300 and the BSDS500.”
An empirical study of context in object detection (2009, methodology): “Pairwise Potential: The pairwise potentials are modeled using probability of boundary (Pb) [25] and probability of occlusion [15] confidences.”
Boundary Detection Using F-Measure-, Filter- and Feature- (F3) Boost (2010, methodology): “This approach has led to improved detectors based on Boosting [6], topological properties of the image [7], spectral gradients [8], multiscale processing [9] and sparse dictionaries [10], among others.”
Exploring Compositional High Order Pattern Potentials for Structured Output Learning (2013, methodology): “Three different ways to define pij yield three pairwise potentials, where the first is image-independent: (1) Set pij to be constant, this would enforce smoothing for the whole image; (2) Set pij to incorporate local contrast information by computing RGB differences between pairs of pixels as in [1]”
Constrained parametric min-cuts for automatic object segmentation (2010, background): “The quality of their algorithm is linked with the excellent globalPb [20] contour detector.”
Contour Detection and Hierarchical Image Segmentation (2011, methodology): “Our gPb detector [3] performs significantly better than other algorithms [2], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28] across almost the entire operating regime.”
Pixel-Level Hand Detection in Ego-centric Videos (2013, methodology): “which uses a combination of video stabilization, gPb [14], super-pixel segmentation and graph cuts to extract hands.”
Object detection and segmentation from joint embedding of parts and pixels (2011, methodology): “As in recent effective image segmentation approaches based on spectral clustering [5, 19, 1], we use the intervening contour cue [10] to define affinities between pixels.”
Recovering Occlusion Boundaries from an Image (2011, methodology): “2004) and Global Pb (Maire et al. 2008) on several external datasets (Sect.”