Implicit Probabilistic Models of Human Motion for Synthesis and Tracking

By Hedvig Kjellström, Michael J. Black, Leonid Sigal. European Conference on Computer Vision, 2002

398 citations recorded by the source corpus.

Subfields: Computational Mechanics, Computer Vision and Pattern Recognition, Control and Systems Engineering

View Paper: https://www.semanticscholar.org/paper/7776d0a6bf3cdf9b3cd18b13d32c6babed84614b

How researchers see this paper

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

In Computational Mechanics

In Computer Vision and Pattern Recognition

In Control and Systems Engineering

Citations in the corpus

Coupled Action Recognition and Pose Estimation from Multiple Views (2012, background): “Approaches can be either on-line, to predict the pose for the next frame (Sidenbladh et al. 2002; Rosenhahn et al. 2007), or offline, to refine the tracked poses (Baak et al.”
HumanEva: Synchronized Video and Motion Capture Dataset for Evaluation of Articulated Human Motion (2006, background): “2000 Sidenbladh [34] Cylinders 2/10 3 T Qualitative 2002 Ronfard [29] Patches 15 2 P Hand Labeled 2002 Sidenbladh [33] Cylinders 2/10 3 T Qualitative 2003 Grauman [9] Mesh N/A 3 P Synthetic/Poser 2003 Ramanan [26] Rectangles 10 2 T,P Hand Labeled 2003 Shakhnarovich [35] Mesh N/A 3 P Synthetic/Poser ”
Incremental learning of full body motion primitives and their sequencing through human motion observation (2012, methodology): “The constructed tree structure bears some similarity to the binary tree proposed by Sidenbladh et al. (2002).”
Efficient Human Pose Estimation from Single Depth Images (2013, background): “Although techniques exist to simulate human motion (e.g., [40]), they do not yet produce a full range of volitional motions of a human subject.”
Seeing Invisible Poses: Estimating 3D Body Pose from Egocentric Video (2016, methodology): “Another implicit model has also been used in third person people tracking [29].”
History Repeats Itself: Human Motion Prediction via Motion Attention (2020, methodology): “history of their previous motion. Predicting human motion can be highly benecial for tasks such as human tracking [6], human-robot interaction [14], and human motion generation for computer graphics [16,15,28]. To tackle the problem eectively, recent approaches use deep neural networks [22,5,11] to ”
Kinematic jump processes for monocular 3D human tracking (2003, methodology): “Sidenbladhet al use particle filtering with importance sampling based on either a learned walki g model or a database of motion snippets, to focus search in the neighborhood of known trajectory pathways [11, 12].”
Tracking People by Learning Their Appearance (2007, background): “The dynamic model PðXtjXt 1Þ can be learned from motion capture data [10], [16] but this often requires choosing the specific motion (walking, running, etc.”
Style-based inverse kinematics (2004, methodology): “Video motion capture using models learned from motion capture data is an active area of research [Brand 1999; Grauman et al. 2003; Howe et al. 2000; Ramanan and Forsyth 2004; Rosales and Sclaroff 2002; Sidenbladh et al. 2002].”
3D People Tracking with Gaussian Process Dynamical Models (2006, background): “It provides a continuous density function over poses and motions that is generally non-Gaussian and multimodal.”
Articulated pose estimation with tiny synthetic videos (2015, background): “Our work requires the first frame to be labeled (a historically commonplace assumption [8, 31, 15, 22]), but differs from past initialized trackers in that no temporal tracking is performed (making our approach resistant to drift).”
Motion capture assisted animation: texturing and synthesis (2002, background): “Such a search could be made more efficient by including probabilistic information about the motion of the subject [47].”
Human motion tracking for rehabilitation - A survey (2008, background): “Camera configuration The ‘‘line of sight’’ problem can be partially tackled using a proper camera setup, including a single camera [6,18,46,123, 142,13,139,140,168,169](see Fig.”
Will the Pedestrian Cross? A Study on Pedestrian Path Prediction (2014, background): “Sidenbladh et al. [15] added an efficient tree search in the context of articulated 3-”
Vision-based human motion analysis: An overview (2007, result): “Sidenbladh et al. [94] take a similar approach.”
3D human pose from silhouettes by relevance vector regression (2004, background): “An important subcase ismodel-based tracking , which focuses on tracking the pose estimate from one time step to the next starting from a known initialization, based on an approximate dynamical model [5, 12].”
Inferring 3D body pose from silhouettes using activity manifold learning (2004, background): “Alternatively, 3D body pose can be directly inferred from the visual input [11, 2, 23, 22, 16, 15, 25].”
Data fusion for visual tracking with particles (2004, neutral): “See examples in [7], [13], [14], [26], [28], [41], [43], [45], [46], [48], and [49].”