A data-driven approach for real-time full body pose reconstruction from a depth camera

By Andreas Baak, Meinard Müller, Gaurav Bharaj, Hans-Peter Seidel, Christian Theobalt. 2011

345 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Endocrinology, Diabetes and Metabolism

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

In Endocrinology, Diabetes and Metabolism

Citations in the corpus

Space-Time Representation of People Based on 3D Skeletal Data: A Review (2016, methodology): “[58] Matching previous joint positions Single depth image 3D skeleton, 20 joints, real-time, 100 fps, robust to sensor noise and occlusions Taylor et al.”
Panoptic Studio: A Massively Multiview System for Social Motion Capture (2015, background): “Depth sensors such as the Kinect [36, 4] are also emerging as a promising sensing modality, yet they suffer from the frequent occlusions of social interaction.”
Panoptic Studio: A Massively Multiview System for Social Interaction Capture (2016, background): “Depth sensors such as the Kinect [39], [40] can produce 3D pose from a single view.”
Real-Time Articulated Hand Pose Estimation Using Semi-supervised Transductive Regression Forests (2013, background): “While latest depth sensor technology has enabled body pose estimation in real-time [2, 24, 12, 26], hand pose estimation still requires improvement.”
Metric Regression Forests for Correspondence Estimation (2015, background): “2010; Baak et al. 2011; Taylor et al. 2012). The most relevant example of such a hybrid system is that of Taylor et al. (2012) who robustly fit a mesh model to a set of image-to-model correspondences predicted by a decision forest.”
Real-Time Human Pose Tracking from Range Data (2012, background): “[21] perform nearest-neighbor search in a database that maps features computed on depth images to poses.”
Total Capture: A 3D Deformation Model for Tracking Faces, Hands, and Bodies (2018, background): “With the introduction of commodity depth sensors, single-view depth-based body motion capture became a popular direction [3, 40].”
Realtime and Robust Hand Tracking from Depth (2014, methodology): “We hope these could advance the experimentation practices for future work.”
Human Action Recognition From Various Data Modalities: A Review (2020, neutral): “[167], [168] and depth data [169], [170], [171], [172], [173], [174].”
The Vitruvian manifold: Inferring dense correspondences for one-shot human pose estimation (2012, methodology): “Using standard linear skinning [4, 3], the limbs allow us to define a mesh model surface.”
Recognizing Actions from Depth Cameras as Weakly Aligned Multi-part Bag-of-Poses (2013, background): “These new acquisition devices have stimulated the development of various promising applications, including human pose reconstruction and estimation [1, 6, 12], scene flow estimation [7], hand gesture recognition [2], face super-resolution [3].”
Efficient ConvNet-based marker-less motion capture in general scenes with a low number of cameras (2015, methodology): “Our approach can also be applied in settings where other approaches for pose estimation with a low number of sensors, that are based on depth cameras [4] or inertial sensors [31], are hard or impossible to be used, e.g. outdoors.”
Parsing the Hand in Depth Images (2014, background): “, the fast point feature histograms [27], the geodesic distance map [33], and the depth kernel descriptors [34].”
Monocular 3D Human Pose Estimation in the Wild Using Improved CNN Supervision (2016, background): “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.”
Monocular Total Capture: Posing Face, Body, and Hands in the Wild (2018, background): “The actively-studied markerless approaches are also based on multi-view systems [21, 29, 19, 25, 26] or depth cameras [50, 7].”
Monocular 3D Human Pose Estimation Using Transfer Learning and Improved CNN Supervision (2016, background): “Special RGB-D cameras enable real-time monocular pose estimation [65] or motion tracking [6], but often do not work in general scenes.”