Filters








606 Hits in 6.4 sec

Probabilistic Character Motion Synthesis using a Hierarchical Deep Latent Variable Model [article]

Saeed Ghorbani, Calden Wloka, Ali Etemad, Marcus A. Brubaker, Nikolaus F. Troje
2020 pre-print
The proposed architecture, which is designed as a hierarchical recurrent model, maps each sub-sequence of motions into a stochastic latent code using a variational autoencoder extended over the temporal  ...  We present a probabilistic framework to generate character animations based on weak control signals, such that the synthesized motions are realistic while retaining the stochastic nature of human movement  ...  Our proposed model for character animation synthesis is based on a deep recurrent neural network.  ... 
doi:10.1111/cgf.14116 arXiv:2010.09950v1 fatcat:fl5iwd5mujhptl3vocutsw3owy

Hierarchical Graph-Convolutional Variational AutoEncoding for Generative Modelling of Human Motion [article]

Anthony Bourached, Robert Gray, Ryan-Rhys Griffiths, Ashwani Jha, Parashkev Nachev
2022 arXiv   pre-print
Here we propose a novel architecture based on hierarchical variational autoencoders and deep graph convolutional neural networks for generating a holistic model of action over multiple time-scales.  ...  Models of human motion commonly focus either on trajectory prediction or action classification but rarely both.  ...  HM-VAE uses a 2-layered hierarchical VAE to learn complex human motions independent of task.  ... 
arXiv:2111.12602v3 fatcat:modbvnwomnd6dhbb7lqzz3phke

A Survey on Deep Learning for Skeleton-Based Human Animation [article]

L. Mourot, L. Hoyet, F. Le Clerc, François Schnitzler
2021 arXiv   pre-print
Second, we cover state-of-the-art approaches divided into three large families of applications in human animation pipelines: motion synthesis, character control and motion editing.  ...  First, we introduce motion data representations, most common human motion datasets and how basic deep models can be enhanced to foster learning of spatial and temporal patterns in motion data.  ...  Fourth, unlike with deep generative models used in motion synthesis, interactivity is at the core of character control with models dynamically reacting to user input flows while trying to provide diverse  ... 
arXiv:2110.06901v1 fatcat:abppln4rbbeufiw4z6a3wnk7oy

Graph-based Normalizing Flow for Human Motion Generation and Reconstruction [article]

Wenjie Yin, Hang Yin, Danica Kragic, Mårten Björkman
2021 arXiv   pre-print
In this paper, we propose a probabilistic generative model to synthesize and reconstruct long horizon motion sequences conditioned on past information and control signals, such as the path along which  ...  We evaluate the models on a mixture of motion capture datasets of human locomotion with foot-step and bone-length analysis.  ...  In [14] , a hierarchical recurrent model is proposed with each motion sub-sequence mapped to a stochastic latent code through a VAE.  ... 
arXiv:2104.03020v1 fatcat:wr5p2zx3j5bkjcfbkvhb6t4n5a

Deep Residual Mixture Models [article]

Perttu Hämäläinen and Martin Trapp and Tuure Saloheimo and Arno Solin
2021 arXiv   pre-print
We propose Deep Residual Mixture Models (DRMMs), a novel deep generative model architecture.  ...  Compared to other deep models, DRMMs allow more flexible conditional sampling: The model can be trained once with all variables, and then used for sampling with arbitrary combinations of conditioning variables  ...  As illustrated in Fig. 3 and detailed in Sec. 3, a characterizing feature of our DRMM architecture is that the output of each layer is a combination of a stochastic latent variable and a modeling residual  ... 
arXiv:2006.12063v3 fatcat:zlvajiiuabazhka3d63wdf4cfq

Transflower: probabilistic autoregressive dance generation with multimodal attention [article]

Guillermo Valle-Pérez, Gustav Eje Henter, Jonas Beskow, André Holzapfel, Pierre-Yves Oudeyer, Simon Alexanderson
2021 arXiv   pre-print
First, we present a novel probabilistic autoregressive architecture that models the distribution over future poses with a normalizing flow conditioned on previous poses as well as music context, using  ...  Using this dataset, we compare our new model against two baselines, via objective metrics and a user study, and show that both the ability to model a probability distribution, as well as being able to  ...  A deep learning framework for Conference on Computer Vision. 5442–5451. character motion synthesis and editing. ACM Trans.  ... 
arXiv:2106.13871v1 fatcat:ofkengygabcfdlbcwsl6vk4rmq

Deep Autoencoding Topic Model with Scalable Hybrid Bayesian Inference [article]

Hao Zhang, Bo Chen, Yulai Cong, Dandan Guo, Hongwei Liu, Mingyuan Zhou
2020 arXiv   pre-print
To build a flexible and interpretable model for document analysis, we develop deep autoencoding topic model (DATM) that uses a hierarchy of gamma distributions to construct its multi-stochastic-layer generative  ...  variational encoder that deterministically propagates information upward via a deep neural network, followed by a Weibull distribution based stochastic downward generative model.  ...  PGBN [17], a deep probabilistic topic model, as the decoder.  ... 
arXiv:2006.08804v1 fatcat:px4gousafnehtf3w55tzeohweu

A Deep Learning Framework for Assessing Physical Rehabilitation Exercises

Yalin Liao, Aleksandar Vakanski, Min Xian
2020 IEEE transactions on neural systems and rehabilitation engineering  
The proposed performance metric is defined based on the log-likelihood of a Gaussian mixture model, and encodes low-dimensional data representation obtained with a deep autoencoder network.  ...  The presented framework is validated using a dataset of ten rehabilitation exercises.  ...  Acknowledgments This work was supported by the Center for Modeling Complex Interactions (CMCI) at the University of Idaho through NIH Award #P20GM104420.  ... 
doi:10.1109/tnsre.2020.2966249 pmid:31940544 pmcid:PMC7032994 fatcat:s5grhmmo6nfcjhto5aj4htm3yy

Spatio-temporal Manifold Learning for Human Motions via Long-horizon Modeling

He Wang, Edmond S. L. Ho, Hubert P. H. Shum, Zhanxing Zhu
2019 IEEE Transactions on Visualization and Computer Graphics  
Recent research has shown that such problems can be approached by learning a natural motion manifold using deep learning on a large amount data, to address the shortcomings of traditional data-driven approaches  ...  In this paper, we propose a new deep network to tackle these challenges by creating a natural motion manifold that is versatile for many applications.  ...  The authors wish to gratefully acknowledge the support of NVIDIA Corporation with the donation of the Titan Xp GPU used for this research.  ... 
doi:10.1109/tvcg.2019.2936810 pmid:31443030 fatcat:ul4scymsbzhh3a6cfg7lqw3erm

Spatio-temporal Manifold Learning for Human Motions via Long-horizon Modeling [article]

He Wang, Edmond S. L. Ho, Hubert P. H. Shum, Zhanxing Zhu
2019 arXiv   pre-print
Recent research has shown that such problems can be approached by learning a natural motion manifold using deep learning to address the shortcomings of traditional data-driven approaches.  ...  In this paper, we propose a new deep network to tackle these challenges by creating a natural motion manifold that is versatile for many applications.  ...  The authors wish to gratefully acknowledge the support of NVIDIA Corporation with the donation of the Titan Xp GPU used for this research.  ... 
arXiv:1908.07214v1 fatcat:cbvk7xhpujeszcm6rapoqg2nzi

Cross-Conditioned Recurrent Networks for Long-Term Synthesis of Inter-Person Human Motion Interactions

Jogendra Nath Kundu, Himanshu Buckchash, Priyanka Mandikal, Rahul M V, Anirudh Jamkhandi, R. Venkatesh Babu
2020 2020 IEEE Winter Conference on Applications of Computer Vision (WACV)  
for long-term generation in a novel hierarchical fashion.  ...  As opposed to prior approaches, we guarantee structural plausibility of 3D pose by training the recurrent model to regress latent representation of a separately trained generative pose embedding network  ...  This work was supported by a Wipro PhD Fellowship (Jogendra), and a project grant from Robert Bosch Centre for Cyber-Physical Systems, IISc.  ... 
doi:10.1109/wacv45572.2020.9093627 dblp:conf/wacv/KunduBMVJR20 fatcat:5oj4z365u5euffic4bu4jkbuje

DEEP DISCRIMINATIVE AND GENERATIVE MODELS FOR SPEECH PATTERN RECOGNITION [chapter]

Li Deng, Navdeep Jaitly
2015 Handbook of Pattern Recognition and Computer Vision  
Both models are characterized as being 'deep' as they use layers of latent or hidden variables.  ...  Understanding and exploiting tradeoffs between deep generative and discriminative models is a fascinating area of research and it forms the background of this chapter.  ...  models, such as DNNs that compute posterior probabilities directly with no probabilistic dependency among latent variables as common in deep generative models.  ... 
doi:10.1142/9789814656535_0002 fatcat:2ovjgqq4njgohffzvdnnut6si4

A Deep Learning Framework for Assessing Physical Rehabilitation Exercises [article]

Y. Liao, A. Vakanski, M. Xian
2019 arXiv   pre-print
The article proposes a new framework for assessment of physical rehabilitation exercises based on a deep learning approach.  ...  Multiple deep network architectures are repurposed for the task in hand and are validated on a dataset of rehabilitation exercises.  ...  Whereas the probabilistic models are advantageous in handling the variability due to the stochastic character of human movements, models with abilities for a hierarchical data representation can produce  ... 
arXiv:1901.10435v2 fatcat:cntjzyj4gjb53kvschfmmu5h4i

A tutorial survey of architectures, algorithms, and applications for deep learning

Li Deng
2014 APSIPA Transactions on Signal and Information Processing  
Using this scheme, I provide a taxonomy-oriented survey on the existing deep architectures and algorithms in the literature, and categorize them into three classes: generative, discriminative, and hybrid  ...  Deep learning refers to a class of machine learning techniques, developed largely since 2006, where many stages of non-linear information processing in hierarchical architectures are exploited for pattern  ...  As the generative component of the DBN, it is a probabilistic model composed of multiple layers of stochastic, latent variables.  ... 
doi:10.1017/atsip.2013.9 fatcat:4l4uonhhcffkbfot2fztpfxo2e

Action2video: Generating Videos of Human 3D Actions [article]

Chuan Guo, Xinxin Zuo, Sen Wang, Xinshuang Liu, Shihao Zou, Minglun Gong, Li Cheng
2021 arXiv   pre-print
Moreover, given an additional input image of a clothed human character, an entire pipeline is proposed to extract his/her 3D detailed shape, and to render in videos the plausible motions from different  ...  Specifically, the Lie algebraic theory is engaged in representing natural human motions following the physical law of human kinematics; a temporal variational auto-encoder (VAE) is developed that encourages  ...  Each learn a θ-parameterized generative model, pθ (x, z), over element of so(3) is in the form of a 3×3 skew-symmetric data x and latent variables z.  ... 
arXiv:2111.06925v2 fatcat:obdpetdfqbdetonei73ndq6ckq
« Previous Showing results 1 — 15 out of 606 results