1,419 Hits in 3.2 sec

ManiFest: Manifold Deformation for Few-shot Image Translation [article]

Fabio Pizzati, Jean-François Lalonde, Raoul de Charette
2022 arXiv   pre-print
We instead propose ManiFest: a framework for few-shot image translation that learns a context-aware representation of a target domain from a few images only.  ...  The learned manifold is interpolated and deformed towards the few-shot target domain via patch-based adversarial and feature statistics alignment losses.  ...  Our approach, illustrated Target (few-shot night) General translation Few-shot learning Baselines translation manifold Deformed Manifold Exemplar translation Fig. 1 : Overview of ManiFest, which  ... 
arXiv:2111.13681v3 fatcat:imw7sp3b5nbyjignhqtjgwzl5u

3D Digital Model of Folk Dance Based on Few-Shot Learning and Gesture Recognition

Ning Zhang, Xin Ning
2022 Computational Intelligence and Neuroscience  
Then, this paper writes the data into the AAM model for 3D digital modeling and retains the information by integrating the manifold ordering.  ...  Finally, this paper designs a folk dance learning method combined with the Few-Shot learning method.  ...  Comparative Analysis of Few-Shot Learning for 3D Digital Modeling of Folk Dance Based on Gesture Recognition e algorithm data set is analyzed through experiments, and finally the Few-Shot learning method  ... 
doi:10.1155/2022/3682261 pmid:35814540 pmcid:PMC9262513 fatcat:sscpavv2jfbzncshjjzi2g2ula

Intrinsic Regularity Detection in 3D Geometry [chapter]

Niloy J. Mitra, Alex Bronstein, Michael Bronstein
2010 Lecture Notes in Computer Science  
Especially challenging is to detect intrinsic regularity, where the repetitions are on an intrinsic grid, without any apparent Euclidean pattern to describe the shape, but rising out of (near) isometric deformation  ...  Acknowledgement We thank Helmut Pottmann and the anonymous reviewers for their comments and helpful suggestions.  ...  In particular, the intrinsic regularity manifests as a grid in the plane.  ... 
doi:10.1007/978-3-642-15558-1_29 fatcat:ik5lwllzojgy5ejw5ik4apxjxu

Geometric Deep Learning: Going beyond Euclidean data

Michael M. Bronstein, Joan Bruna, Yann LeCun, Arthur Szlam, Pierre Vandergheynst
2017 IEEE Signal Processing Magazine  
Geometric deep learning is an umbrella term for emerging techniques attempting to generalize (structured) deep neural models to non-Euclidean domains such as graphs and manifolds.  ...  Some examples include social networks in computational social sciences, sensor networks in communications, functional networks in brain imaging, regulatory networks in genetics, and meshed surfaces in  ...  In tasks that are translation invariant we have |y(L τ f ) − y(f )| ≈ ∇τ , (3) for all f, τ . Here, ∇τ measures the smoothness of a given deformation field.  ... 
doi:10.1109/msp.2017.2693418 fatcat:ppyimvcbwfd55lsc5mkcmsy7ai

Point Set Registration Using Havrda–Charvat–Tsallis Entropy Measures

Nicholas J Tustison, Suyash P Awate, Gang Song, Tessa S Cook, James C Gee
2011 IEEE Transactions on Medical Imaging  
A variant of the traditional free-form deformation approach, known as directly manipulated free-form deformation, is used to model the transformation of the registration solution.  ...  Characterization of the proposed framework includes comparison with other state of the art kernel-based methods and demonstration of its utility for lung registration via labeled point set representation  ...  Altes at the University of Virginia for providing the hyperpolarized image data.  ... 
doi:10.1109/tmi.2010.2086065 pmid:20937578 fatcat:kzo5ub3yyncavagylxpruncema

Synchronised photoreversion of spirooxazine ring opening in thin crystals to uncover ultrafast dynamics

Khalid M. Siddiqui, Gastón Corthey, Stuart A. Hayes, Andreas Rossos, Daniel S. Badali, Rui Xian, R. Scott Murphy, Benjamin J. Whitaker, R. J. Dwayne Miller
2016 CrysteEngComm  
Acknowledgements This work was funded by the Max Planck Society in collaboration with the Centre for Free Electron Laser Science and the Hamburg Centre for Ultrafast Imaging. K. M.  ...  Kochman for useful discussions.  ...  On the other hand, for systems in the solid state this is not possible and, therefore, samples are either translated or replaced after a few laser shots.  ... 
doi:10.1039/c6ce01049k fatcat:khcecznscrdo5a3hurmd4hxp6a

Geometric Deep Learning: Grids, Groups, Graphs, Geodesics, and Gauges [article]

Michael M. Bronstein, Joan Bruna, Taco Cohen, Petar Veličković
2021 arXiv   pre-print
learning is built from two simple algorithmic principles: first, the notion of representation or feature learning, whereby adapted, often hierarchical, features capture the appropriate notion of regularity for  ...  We hope that our perspective will make it easier both for newcomers and practitioners to navigate the field, and for researchers to synthesise novel architectures, as instances of our blueprint.  ...  For two-dimensional manifolds (surfaces), isometries can be intuitively understood as inelastic deformations that deform the manifold without 'stretching' or 'tearing' it.  ... 
arXiv:2104.13478v2 fatcat:odbzfsau6bbwbhulc233cfsrom

The Power of Points for Modeling Humans in Clothing [article]

Qianli Ma and Jinlong Yang and Siyu Tang and Michael J. Black
2021 arXiv   pre-print
The code is available for research purposes.  ...  The network is trained from 3D point clouds of many types of clothing, on many bodies, in many poses, and learns to model pose-dependent clothing deformations.  ...  creation from a "one-shot" observation.  ... 
arXiv:2109.01137v2 fatcat:c2grh7j3qbesdbt4susat3vecy

Shape google

Alexander M. Bronstein, Michael M. Bronstein, Leonidas J. Guibas, Maks Ovsjanikov
2011 ACM Transactions on Graphics  
The computer vision and pattern recognition communities have recently witnessed a surge of feature-based methods in object recognition and image retrieval applications.  ...  These methods allow representing images as collections of "visual words" and treat them using text search approaches following the "bag of features" paradigm.  ...  Acknowledgment We are grateful to Zhouhui Lian and Umberto Castellani for providing the performance of their algorithms on the SHREC'10 benchmark, and to Giuseppe Patané for providing the FEM computation  ... 
doi:10.1145/1899404.1899405 fatcat:rwtead35svcmrplfxr3lfmtz6a

StyleMeUp: Towards Style-Agnostic Sketch-Based Image Retrieval [article]

Aneeshan Sain, Ayan Kumar Bhunia, Yongxin Yang, Tao Xiang, Yi-Zhe Song
2021 arXiv   pre-print
Sketch-based image retrieval (SBIR) is a cross-modal matching problem which is typically solved by learning a joint embedding space where the semantic content shared between photo and sketch modalities  ...  Extensive experiments show that our style-agnostic model yields state-of-the-art performance for both category-level and instance-level SBIR.  ...  Most existing meta-learning methods [54, 46, 6] are designed for few-shot image classification and thus are suitable for our problem of meta-learning of a generalisable cross-modal disentanglement model  ... 
arXiv:2103.15706v2 fatcat:ukbeu2bpujb53j3pzd5zbdldai

Statistical Methods and Models for Video-Based Tracking, Modeling, and Recognition

Rama Chellappa
2009 Foundations and Trends® in Signal Processing  
Central Projection Central projection is the fundamental principle behind imaging with a pinhole camera, and it serves as a good approximation for lens-based imaging for the applications considered here  ...  We will explore the fundamental connections between these different 14 Geometric Models for Imaging highlights are accounted using the specular term of the Phong model.  ...  The third author wishes to thank the members of the imaging group at MERL for their support.  ... 
doi:10.1561/2000000007 fatcat:o5hmdnzbqvbdzjdu72jkojl5ya

Pose Induction for Novel Object Categories

Shubham Tulsiani, Joao Carreira, Jitendra Malik
2015 2015 IEEE International Conference on Computer Vision (ICCV)  
We address the task of predicting pose for objects of unannotated object categories from a small seed set of annotated object classes.  ...  We also show qualitative results on a diverse set of classes and further demonstrate the applicability of our system for learning shape models of novel object classes.  ...  We gratefully acknowledge NVIDIA corporation for the donation of Tesla GPUs for this research.  ... 
doi:10.1109/iccv.2015.16 dblp:conf/iccv/TulsianiCM15 fatcat:6nrj6gnxu5cjfnk3qhelmcjb2u

Stable Semi-local Features for Non-rigid Shapes [chapter]

Roee Litman, Alexander M. Bronstein, Michael M. Bronstein
2012 Mathematics and Visualization  
is presented, which can be used for geometric feature detection in deformable shapes.  ...  Following the success of this approach in image analysis, there is a growing interest in finding analogous methods in the 3D world.  ...  Acknowledgment We are grateful to Dan Raviv for providing us his volume rasterization and Laplacian disretization code. M. M.  ... 
doi:10.1007/978-3-642-34141-0_8 fatcat:qndwdyurbfe4tjbplf323h2ie4

Pose Induction for Novel Object Categories [article]

Shubham Tulsiani, João Carreira, Jitendra Malik
2015 arXiv   pre-print
We address the task of predicting pose for objects of unannotated object categories from a small seed set of annotated object classes.  ...  We also show qualitative results on a diverse set of classes and further demonstrate the applicability of our system for learning shape models of novel object classes.  ...  We gratefully acknowledge NVIDIA corporation for the donation of Tesla GPUs for this research.  ... 
arXiv:1505.00066v2 fatcat:inxkvsrl6renxnvmk7icvhxsya

Video motion analysis for the synthesis of dynamic cues and Futurist art

J.P. Collomosse, P.M. Hall
2006 Graphical Models  
This paper presents new methods for stylising video to produce cartoon motion emphasis cues and modern art.  ...  We describe methods for automatically synthesising such cues within video premised upon the recovery of articulated figures, and the subsequent manipulation of the recovered pose trajectories.  ...  An eigenmodel is generated from mug-shots of many people by considering each image as a vector in some high-dimensional space.  ... 
doi:10.1016/j.gmod.2006.05.003 fatcat:amd4zqlyavalxf7o7c5emwjt54
« Previous Showing results 1 — 15 out of 1,419 results