Segmenting Motion Capture Data into Distinct Behaviors

By Jernej Barbic, Alla Safonova, Jia-Yu Pan, Christos Faloutsos, Jessica K. Hodgins, Nancy S. Pollard. Graphics Interface, 2004

496 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/2080dde9d4a1c196eca6f135bf2eeeb771f7fae5

How researchers see this paper

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

Citations in the corpus

Hierarchical Aligned Cluster Analysis for Temporal Clustering of Human Motion (2013, methodology): “To temporally smooth the time series, following [15] we removed short segments as follows: Suppose that we are given a sequence X 2 IR d (cid”
Temporal segmentation and activity classification from first-person sensing (2009, methodology): “Unsupervised techniques for learning motion primitives from data have recently drawn the interest of many scientists in computer vision [9], [28], [27], [17] and computer graphics [4], [16], [3], [8].”
Ongoing human action recognition with motion capture (2014, methodology): “[4] have used the Principal Component Analysis (PCA) to extract a known number of similar behaviors.”
Structured Time Series Analysis for Human Action Segmentation and Recognition (2014, methodology): “It is notable that, as a kernelized learning algorithm, KTC can be applied for structured sequential data in general, such as multivariate time series and dynamic graphs.”
Segmenting sign language into motor primitives with Bayesian binning (2013, methodology): “Three methods for motion capture data segmentation are compared in Barbič et al. (2004): segment-wise PCA, probabilistic PCA (pPCA) and finite Gaussian mixture models.”
SWIRL: A sequential windowed inverse reinforcement learning algorithm for robot tasks with delayed rewards (2018, methodology): “Often these segments are manually designed or derived from a dictionary of motion primitives, but recently, there are several approaches for learning segmentation criteria automatically by identifying locally similar structure in demonstration data (Barbič et al. 2004; Chiappa and Peters 2010; Alvar”
Effective Active Skeleton Representation for Low Latency Human Action Recognition (2016, methodology): “[46] decomposed motion sequences into distinct action segments by extending probabilistic principal analysis for change-point detection and Gong et al.”
Integrating Driving Behavior and Traffic Context Through Signal Symbolization for Data Reduction and Risky Lane Change Detection (2018, methodology): “in continuous data [21], [22], focusing on local dynamics and the predictability of motion [23]–[27], utilizing short-term con-”
Action recognition on motion capture data using a dynemes and forward differences representation (2014, methodology): “[30] proposed three methods for automatic action segmentation on a motion capture sequence.”
Berkeley MHAD: A comprehensive Multimodal Human Action Database (2013, background): “The spatial segmentation and labeling of mocap data with respect to corresponding body segments is usually trivial; however, as in video, the automatic temporal segmentation of longer sequences into low-level primitives is still an open problem [4] [19].”
Manifold Warp Segmentation of Human Action (2018, methodology): “based on the assumption that Gaussian models of two separate behaviors will be quite different [10].”
Sharing Features among Dynamical Systems with Beta Processes (2009, methodology): “Consider a probability space Θ, and let B 0 denote a finite base measure on Θ with total mass B 0 (Θ) = α. Assuming B 0 is absolutely continuous, we define the following Lévy measure on the product space [0] [1] × Θ: ν(dω, dθ) = cω −1 (1 − ω) c−1 dωB 0 (dθ).”
Sequence of the Most Informative Joints (SMIJ): A new representation for human skeletal action recognition (2012, background): “007 the principle component analysis and data compression theory [20–22].”
Combined shape analysis of human poses and motion units for action segmentation and recognition (2015, unknown): “For instance, in [1] a method based on probabilistic principal component analysis is proposed, which detects when the distribution of human poses change over time.”