Covariance scaled sampling for monocular 3D body tracking

By Cristian Sminchisescu, Bill Triggs. Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. CVPR 2001, 2001

437 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/2d0b50d3df26b64ec5a2f949bc241b4fce515fa9

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

Looking at People: Sensing for Ubiquitous and Wearable Computing (2000, methodology): “The CalTech arm-tracking system [102] tracks upper and lower arm in real-time using a single camera and a standard computer.”
Data fusion for visual tracking with particles (2004, neutral): “See examples in [7], [13], [14], [26], [28], [41], [43], [45], [46], [48], and [49].”
Accurate 3D pose estimation from a single depth image (2011, methodology): “While the prevailing methods in marker-less mocap require a surrounding camera array, solutions using a single video camera have been explored [6, 22].”
Monocular 3D Human Pose Estimation Using Transfer Learning and Improved CNN Supervision (2016, background): “Model-based and multi-view approaches Our problem relates to marker-less motion capture [45, 63] that tracks 3D pose by fitting a model to monocular or multi-view video, often under controlled conditions [83, 71, 72, 80, 7, 24, 73].”
Kernel-Based Object Tracking (2003, neutral): “Explicit tracking approaches of people [69] are timeconsuming and often the simpler blob model [75] or adaptive mixture models [53] are also employed.”
Tracking People by Learning Their Appearance (2007, neutral): “One solution is to intelligently resample; strategies include annealing [12] and covariance-scaled sampling [14].”
Human3.6M: Large Scale Datasets and Predictive Methods for 3D Human Sensing in Natural Environments (2014, neutral): “matic discriminative prediction [10], [11], [12], [13], [14], whereas others aim at model-image alignment [15], [16], [17], [18], [19], [20], [21], [22], [23] or accurate modeling of 3D shape[24], [25] or clothing[26].”
Reconstruction of articulated objects from point correspondences in a single uncalibrated image (2000, result): “This ambiguity is similar to the ambiguities in positioning the segments of the model described by Lee and Chen [9] and Goncalves Bernardo Ursella and Perona [5].”
MoDeep: A Deep Learning Framework Using Motion Features for Human Pose Estimation (2014, positive): “Several other model based articulated tracking systems have been reported over the past two decades, most notably [10, 11, 12, 13, 14, 15, 16].”
Keep It SMPL: Automatic Estimation of 3D Human Pose and Shape from a Single Image (2016, neutral): “While previous work uses approximations to detect interpenetrations [38,46], we believe this regression from shape parameters is novel.”
Monocular 3D Human Pose Estimation in the Wild Using Improved CNN Supervision (2016, neutral): “Unlike marker-less 3D motion capture methods that track articulated human poses from multi-view video sequences, [70, 56, 57, 67, 5, 17, 58, 11] or use active RGB-D cameras [52, 4], our approach is designed to work from a single low-cost RGB camera.”
Kinematic jump processes for monocular 3D human tracking (2003, background): “It is important to realize that the principal axes of the covariance change drastically depending on the viewing direction, and that even if these are computed and used for sampling (as in [14]), they are only local measures that capture little information about the global minimum structure.”
Stochastic Tracking of 3D Human Figures Using 2D Image Motion (2000, neutral): “Methods for full body tracking typically use simple cues such as background difference images [4], color [22] or edges [7,9,10,15].”
Survey on Emotional Body Gesture Recognition (2018, methodology): “In case of some systems, the input data were obtained using a single camera [100], [101].”
Cascaded pose regression (2010, methodology): “[18] computed 3D arm pose iteratively, refining 2D image features with each improved pose estimate; more recently, Ramanan [26] used similar ideas in a multi-stage pose estimation procedure.”
Tracking loose-limbed people (2004, neutral): “5D [17], or 3D [2, 5, 21, 23] leading to a high-dimensional parameter space (25–50 dimensions is not uncommon).”
Recent developments in human motion analysis (2003, background): “As far as tracked objects are concerned, tracking may be classified into tracking of human body parts such as hand, face, and leg [9-12,21-22,57-61] and tracking of whole body [62-97].”
Implicit Probabilistic Models of Human Motion for Synthesis and Tracking (2002, neutral): “See [38] for recent work on human motion tracking that places particles more effectively thus allowing more general motion priors.”
Articulated body motion capture by annealed particle filtering (2000, background): “Goncalves [6] and Deutscher [3] assume a constant angle of view of the subject as does Bregler [2] and Rehg [16].”