Learning to parse images of articulated bodies

By Deva Ramanan. Neural Information Processing Systems, 2006

514 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/6dd0597f8513dc100cd0bc1b493768cde45098a9

How researchers see this paper

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

In Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Articulated part-based model for joint object detection and pose estimation (2011, methodology): “5(a,b) shows that our object detection performance is slightly better than [3] (which achieves the best performance on PASCAL 2010 - human category) on poselet dataset [4] but signiﬁcantly out-performs [3] on [15], respectively.”
3D Human pose estimation: A review of the literature and analysis of covariates (2016, methodology): “For the 2D pose estimation task we employed the approach of Yang and Ramanan [61], which uses a pre-trained 2D human skeletal model trained on the PARSE dataset [160].”
Learning shape models for monocular human pose estimation from the Microsoft Xbox Kinect (2011, methodology): “In contrast to methods using explicit models of limb shape, much recent work has adopted a discriminative approach, effectively using “sliding-window” limb/non-limb classifiers [18, 1, 14, 4].”
Attribute And-Or Grammar for Joint Parsing of Human Attributes, Part and Pose (2016, methodology): “Some recent works use a pre-trained pose detection module [4] as a pre-processing stage and then predict attributes sequentially.”
Clustered Pose and Nonlinear Appearance Models for Human Pose Estimation (2010, methodology): “State-of-the-art methods typically use graphical models which represent the human body as a connected collection of parts corresponding to head, torso and other limbs [1] [6] [9] [12] [16].”
Articulated Human Detection with Flexible Mixtures of Parts (2013, methodology): “Most human pose datasets include images with labeled joint positions [9], [7], [3].”
Articulated Pose Estimation by a Graphical Model with Image Dependent Pairwise Relations (2014, background): “Now we consider the problem of learning the model parameters from images with labeled part locations, which is the data available in most of the human pose datasets [17, 7, 10, 20].”
Articulated Pose Estimation Using Discriminative Armlet Classifiers (2013, methodology): “We might aim to do so by fitting a stick figure model in one of its numerous manifestations [1] [7] [9] [19] [20] by detecting rectangles in the upper and lower parts of the arms.”
Detecting Actions, Poses, and Objects with Relational Phraselets (2012, methodology): “To evaluate this, we trained and evaluated our model on the PARSE benchmark [21].”
Zero-Shot Learning via Visual Abstraction (2014, methodology): “We also use a subset of the standard PARSE [21] dataset, which originally contains 305 images of individuals in various poses.”
Learning hierarchical poselets for human parsing (2011, methodology): “[1], and the iterative parsing method by Ramanan [16].”
Sparse Representation for 3D Shape Estimation: A Convex Relaxation Approach (2015, methodology): “To illustrate the real applicability of the proposed method, we apply the 2D pose detector developed by Yang and Ramanan [54] on the PARSE dataset [55] and then use the proposed method to lift the detected 2D poses to 3D poses with the dictionary learned on the CMU dataset.”
Guiding model search using segmentation (2005, methodology): “We test our algorithm on the people dataset used in previous work [15, 17].”
A Hierarchical Model of Shape and Appearance for Human Action Classification (2007, background): “Generally speaking, there are three popular types of features: static features based on edges and limb shapes [7, 11, 15]; dynamic features based on optical flows [7, 9, 18], and spatialtemporal features that characterizes a space-time volume of the data [2, 6, 8, 13].”
Better Appearance Models for Pictorial Structures (2009, methodology): “By estimating left/right arm parts together we increase the number of training examples (this exploits the appearance similarity of symmetric parts, as done in [11, 15, 21]).”