Priors for people tracking from small training sets

By Raquel Urtasun, David J. Fleet, Aaron Hertzmann, Pascal Fua. Tenth IEEE International Conference on Computer Vision (ICCV'05) Volume 1, 2005

304 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Control and Systems Engineering, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/7c15fe581f5521b16f04258689f8622b3a0019bd

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 Control and Systems Engineering

In Statistics and Probability

Citations in the corpus

Active Learning with Gaussian Processes for Object Categorization (2007, background): “Handling multiple objects in the same image is an interesting and more challenging problem and will be the focus of future work.”
Vision-based human motion analysis: An overview (2007, methodology): “[116] use a GPLVM to learn prior models for 3D human tracking.”
Dynamical binary latent variable models for 3D human pose tracking (2010, methodology): “Perhaps the most prominent current latent variable models are derived from the Gaussian Process Latent Variable Model [8] [23] and the Gaussian Process Dynamical Model [24].”
Learning for Larger Datasets with the Gaussian Process Latent Variable Model (2007, methodology): “It has been succesfully applied in a range of application domains including graphics [Grochow et al., 2004] and visual tracking [Urtasun et al., 2005].”
Coupled Action Recognition and Pose Estimation from Multiple Views (2012, background): “Among the simplest are those based on PCA (Baumberg and Hogg 1994; Sidenbladh et al. 2000; Urtasun et al. 2005).”
Learning Joint Top-Down and Bottom-up Processes for 3D Visual Inference (2006, methodology): “Generative human models and efficient inference algorithms are described in [6, 24, 25, 26, 30 ].”
Learning to Recognize Objects from Unseen Modalities (2010, methodology): “To overcome this issue, we rely on Gaussian processes (GPs) which have proven effective when trained from a small number of examples [15].”
Gaussian Processes for Object Categorization (2010, methodology): “While they have been used in [41, 42] for human motion modeling and in [48] for stereo segmentation, we are unaware of any prior work on visual object recognition in a Gaussian Process framework.”
Structured Prediction of 3D Human Pose with Deep Neural Networks (2016, methodology): “This has been done by explicitly enforcing physical constraints at test time [25, 29] and by data-driven priors over the pose space [6, 28, 33].”
Retrieving actions in movies (2007, background): “3D human tracking and 3D action recognition [22, 26] usually rely on the available tracks of body parts.”
Iterated Second-Order Label Sensitive Pooling for 3D Human Pose Estimation (2014, background): “Many of the early approaches to 3D human pose estimation employed generative models to search the state space in order to identify configurations of a body model that would best align with image features[12, 32, 29, 34, 15].”
A Simple, Fast and Highly-Accurate Algorithm to Recover 3D Shape from 2D Landmarks on a Single Image (2016, background): “Methods such as [20], [21], [22], [23] fall into the category of nonlinear mapping, but like some of the aforementioned algorithms, they are limited to certain types of object categories, e.g., [3] is designed for reconstructing 3D human pose from 2D image landmarks and [4] focuses on 3D car…”
3D People Tracking with Gaussian Process Dynamical Models (2006, background): “Estimates are obtained with deterministic optimization, and look remarkably good despite very noisy, missing or erroneous image data and significant occlusions.”
Local distance preservation in the GP-LVM through back constraints (2006, methodology): “Indeed the successful applications and extensions of the GP-LVM (Grochow et al., 2004; Urtasun et al., 2005; Shon et al., 2006; Wang et al., 2006) each take explicit advantage of the fact that the model is probabilistic.”
Hierarchical Gaussian process latent variable models (2007, methodology): “, 2004) learning a prior model for tracking (Urtasun et al., 2005; Urtasun et al., 2006) and robot simultaneous localisation and mapping (Ferris et al.”