Efficient matching of pictorial structures

By Pedro F. Felzenszwalb, Dan Huttenlocher. Proceedings IEEE Conference on Computer Vision and Pattern Recognition. CVPR 2000 (Cat. No.PR00662), 2000

446 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/6deeed19c56ec6e737a909de9ee172b93f0d0a89

How researchers see this paper

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

In Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

Learning effective human pose estimation from inaccurate annotation (2011, methodology): “In this section we describe the basis of our method and propose extensions to the state-of-the-art clustered PSM approach [11, 17] which improve both the accuracy and efficiency of body part appearance models and handle training using poses with incomplete part visibility.”
Mixtures of trees for object recognition (2001, methodology): “[1, 3]) and have been applied to faces [10, 13, 14].”
MAP Estimation of Semi-Metric MRFs via Hierarchical Graph Cuts (2009, background): “Although these may seem like very restrictive assumptions, several problems in computer vision and related areas can be expressed using semi-metric labeling, from low level tasks like image denoising and stereo reconstruction [30] to high level tasks like pose estimation [9] and scene segmentation [”
Human Pose Estimation Using a Joint Pixel-wise and Part-wise Formulation (2013, methodology): “Others have proposed modelling dependencies and relationships between multiple people [9], which addresses problem (a), and methods for efficiently sampling from pictorial structures [6, 11, 19].”
Strike a pose: tracking people by finding stylized poses (2005, methodology): “Given an edge image on the left, we search for a tree pictorial structure [6] using rectangle”
The Fastest Pedestrian Detector in the West (2010, background): “Through numerous strategies, including cascades [28], coarse-to-fine search [17], distance transforms [13], branch and bound search [19], and many others, the classification stage can be made quite fast.”
Unsupervised Learning of Object Deformation Models (2007, methodology): “As in [7], the Minimum Spanning Tree algorithm is used to construct a tree-structured graph connecting all MRF nodes.”
Human Detection Based on a Probabilistic Assembly of Robust Part Detectors (2004, background): “Similarly, Felzenszwalb and Huttenlocher [2] show that dynamic programing can be used to group body plans efficiently, but simplistic colour-based part detectors are applied.”
OBJ CUT (2005, methodology): “Given an image D containing an instance of a known object category, e.g. cows, we wish to segment the image into figure, i.e. pixels belonging to the object, and ground, i.e. the background.”
Crosstalk Cascades for Frame-Rate Pedestrian Detection (2012, background): “Finally, research on optimizing other types of classifiers [1,2,3,17,18,19] is of considerable interest, however, such methods have difficulty matching the speed achieved by boosted classifiers.”
Multiclass recognition and part localization with humans in the loop (2011, methodology): “Specifically, we use a part representation similar to a Felzenszwalb-style deformable part model [8, 9] (sliding window HOG-based part detectors fused with treestructured spatial dependencies).”
Structured Learning and Prediction in Computer Vision (2011, methodology): “8 Pictorial structure model of Felzenszwalb and Huttenlocher [37] for person recognition.”
Exploring the Spatial Hierarchy of Mixture Models for Human Pose Estimation (2012, background): “A tree model [5,8,17] yields fast and exact inference, but it may not be able to encode sufficient constraints to avoid problems such as double-counting arms and legs.”
The Ignorant Led by the Blind: A Hybrid Human–Machine Vision System for Fine-Grained Categorization (2014, methodology): “Since the datasets that we use label part locations (x p, yp) but not aspects vp, aspect labels are assigned prior to training using pose clustering techniques—this practice is widely used for popular implementations of DPMs (Yang and Ramanan 2011) and poselets (Bourdev and Malik 2009).”
Beyond trees: common-factor models for 2D human pose recovery (2005, methodology): “For the treestructured graph, T , the MAP estimation problem can be solved in O(nh) time, where n is the number of parts and h is the number of possible locations of each part (using the methods in [4]).”
Cascaded pose regression (2010, unknown): “Although this generates a large number of tests, sliding window methods can be made more efficient through cascades [28], distance transforms [13], branch-and-bound search [20] and coarse to fine approaches [15].”
Segmentation using Deformable Spatial Priors with Application to Clothing (2010, unknown): “The idea is to augment the posterior probability to be maximized with an additional term that reflects the conformity of the labelling to a prior shape model - in this case a (layered) pictorial structure [9, 12].”