Filters








22 Hits in 7.3 sec

Segmentation and Recovery of Superquadric Models using Convolutional Neural Networks [article]

Jaka Šircelj, Tim Oblak, Klemen Grm, Uroš Petković, Aleš Jaklič, Peter Peer, Vitomir Štruc, Franc Solina
2020 arXiv   pre-print
Specifically, we present a (two-stage) approach built around convolutional neural networks (CNNs) capable of segmenting complex depth scenes into the simpler geometric structures that can be represented  ...  In the first stage, our approach uses a Mask RCNN model to identify superquadric-like structures in depth scenes and then fits superquadric models to the segmented structures using a specially designed  ...  Segment-then-fit. Techniques from this group follow a two-stage procedure, where the data is first segmented up front and independently of superquadric recovery [10] .  ... 
arXiv:2001.10504v1 fatcat:bjvhv6gpzrfd3gfpnpfpehh3mq

Learning to predict superquadric parameters from depth images with explicit and implicit supervision

Tim Oblak, Jaka Sircelj, Vitomir Struc, Peter Peer, Franc Solina, Ales Jaklic
2020 IEEE Access  
We present a new learning objective that relies on the superquadric (inside-outside) function and develop two learning strategies for training convolutional neural networks (CNN) capable of predicting  ...  Reconstruction of 3D space from visual data has always been a significant challenge in the field of computer vision.  ...  In 1987, Solina and Bajcsy [23] devised a least-squares minimization process for superquadric recovery from range images, which they further refined a few years later [9] .  ... 
doi:10.1109/access.2020.3041584 fatcat:pvatbaixb5aabmvy4i4n3j7y3i

A cognitive architecture for artificial vision

A. Chella, M. Frixione, S. Gaglio
1997 Artificial Intelligence  
A new cognitive architecture for artificial vision is proposed.  ...  This link is modeled as a time delay attractor neural network. Results are reported obtained by an experimental implementation of the architecture.  ...  Acknowledgements We would like to thank Luigia Carlucci Aiello and Peter Gardenfors for the interesting discussions about the topics of the paper.  ... 
doi:10.1016/s0004-3702(96)00039-2 fatcat:lmot3ytbivbi3l4pxcjt7fckxa

Cuboids Revisited: Learning Robust 3D Shape Fitting to Single RGB Images [article]

Florian Kluger, Hanno Ackermann, Eric Brachmann, Michael Ying Yang, Bodo Rosenhahn
2021 arXiv   pre-print
A RANSAC estimator guided by a neural network fits these primitives to 3D features, such as a depth map.  ...  Previous approaches directly estimate shape parameters from a 2D or 3D input, and are only able to reproduce simple objects, yet unable to accurately parse more complex 3D scenes.  ...  This estimator predicts sampling weights p = f w (Y, s) from observations Y and a state s via a neural network with parameters w, which are learnt from data.  ... 
arXiv:2105.02047v1 fatcat:onipwd2ikrbqxfoa4thp4gusdy

The Visual Analysis of Human Movement: A Survey

D.M Gavrila
1999 Computer Vision and Image Understanding  
The ability to recognize humans and their activities by vision is key for a machine to interact intelligently and effortlessly with a human-inhabited environment.  ...  This survey identifies a number of promising applications and provides an overview of recent developments in this domain.  ...  In other work, Pentland [62] fit deformable superquadrics to range data. A maximum-likelihood technique provides the initial part segmentation based on the object silhouette.  ... 
doi:10.1006/cviu.1998.0716 fatcat:ep75ekg4h5elzebmm3pasqlqdy

Diffusion MRI in the heart

Choukri Mekkaoui, Timothy G. Reese, Marcel P. Jackowski, Himanshu Bhat, David E. Sosnovik
2015 NMR in Biomedicine  
In this review, we describe the solutions and approaches that have been developed to enable diffusion MRI of the heart in vivo, including a dual-gated stimulated echo approach, a velocity-(M 1 ) or an  ...  Diffusion MRI provides unique information on the structure, organization, and integrity of the myocardium without the need for exogenous contrast agents.  ...  Acknowledgements This work was supported by the following grants from the National Institutes of Health: R01HL093038 (DES), R01HL112831 (DES), and P41RR14075 (Athinoula A.  ... 
doi:10.1002/nbm.3426 pmid:26484848 pmcid:PMC5333463 fatcat:msn7zambgngxhgbl27brlpe6z4

A surface-based technique for warping three-dimensional images of the brain

P. Thompson, A.W. Toga
1996 IEEE Transactions on Medical Imaging  
Nevertheless, no two people's brains are the same, and this presents a challenge for any attempts to create standardized atlases.  ...  Such normal variations have severely complicated the goals of comparing functional and anatomic data from many subjects, and of developing standardized atlases of the human brain.  ...  .), by a Pre-Doctoral Fellowship of the Howard Hughes Medical Institute, and by a Fulbright Scholarship from the U.S.-U.K. Fulbright Commission, London (P.T.).  ... 
doi:10.1109/42.511745 pmid:18215923 fatcat:gip2kiwvrrgyfhqin3nadeksxi

A Survey Of Free-Form Object Representation and Recognition Techniques

Richard J. Campbell, Patrick J. Flynn
2001 Computer Vision and Image Understanding  
This survey reviews recent literature on both the 3D model building process and techniques used to match and identify free-form objects from imagery.  ...  The availability of detailed data describing the shape of an object offers the computer vision practitioner new ways to recognize and localize free-form objects.  ...  The authors thank three anonymous reviewers and Kevin Bowyer for helpful comments and suggestions.  ... 
doi:10.1006/cviu.2000.0889 fatcat:lujlciputzdb7e3pjhm4sfknfy

Recovering 3D Human Mesh from Monocular Images: A Survey [article]

Yating Tian, Hongwen Zhang, Yebin Liu, Limin Wang
2022 arXiv   pre-print
Meanwhile, continuous efforts are devoted to improving the quality of 3D mesh labels for a wide range of datasets.  ...  Estimating human pose and shape from monocular images is a long-standing problem in computer vision.  ...  [143] explore a R-CNN-based architecture for detection and estimation for all people in the image.  ... 
arXiv:2203.01923v2 fatcat:vb6xa5wdsrhdxd2ebvg54qq2m4

Point2Cyl: Reverse Engineering 3D Objects from Point Clouds to Extrusion Cylinders [article]

Mikaela Angelina Uy, Yen-yu Chang, Minhyuk Sung, Purvi Goel, Joseph Lambourne, Tolga Birdal, Leonidas Guibas
2021 arXiv   pre-print
Reverse engineering from a raw geometry to a CAD model is an essential task to enable manipulation of the 3D data in shape editing software and thus expand their usages in many downstream applications.  ...  Precisely, our approach first predicts per-point segmentation, base/barrel labels and normals, then estimates for the underlying extrusion parameters in differentiable and closed-form formulations.  ...  recovery of extrusion axis from to B̂:,0 == 0 for barrel points, B̂:,1 == 1 for the base, and points).  ... 
arXiv:2112.09329v1 fatcat:j2ifiimyebcz7ccw6vslx5pvfa

Towards Building and Civil Infrastructure Reconstruction from Point Clouds: A Review on Data and Key Techniques

Yusheng Xu, Uwe Stilla
2021 IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing  
There are few systematic reviews summarizing the ways of acquiring 3-D points and the techniques for reconstructing 3-D objects from point clouds for application scenarios in a built environment or construction  ...  The reconstruction of 3-D objects in a scene of man-made infrastructure and buildings is one of the core tasks using point clouds, which involves both the 3-D data acquisition and processing.  ...  For instance, in [20] , the octree architecture and the region growing frame are combined for rapid surface patch segmentation.  ... 
doi:10.1109/jstars.2021.3060568 fatcat:zjpzyfmibjeghcf35csjnwuseu

A posture sequence learning system for an anthropomorphic robotic hand

Antonio Chella, Haris Džindo, Ignazio Infantino, Irene Macaluso
2004 Robotics and Autonomous Systems  
These are the need to define a metric of imitation performance and the need to determine a representation common to visual and motor systems.  ...  Work in that area tackles the development of robust algorithms for motor control, motor learning, gesture recognition and visuo-motor integration.  ...  Acknowledgments This work was partially funded by the ROBEA-CNRS project "Modeles Bayesiens pour la Generation de Mouvement", the BIBA project funded by the European Community, and a grant from the French  ... 
doi:10.1016/j.robot.2004.03.008 fatcat:fefnr5l7gnhnhaca5nw52f3kru

Vision and action

Cornelia Fermüller, Yiannis Aloimonos
1995 Image and Vision Computing  
Our suggested approach for understanding behavioural vision to realize the relationships of perception and action builds on two earlier approaches, the Medusa philosophy' and the Synthetic approach*.  ...  In this paper, we present our views about the architecture of vision systems, about how to tackle the design and analysis of perceptual systems, and promising future research directions.  ...  The route consists of path segments, each of which takes the animal for a given distance in a given direction.  ... 
doi:10.1016/0262-8856(95)98754-h fatcat:62wpierdfnapbmohs2apohp5la

A Novice Guide towards Human Motion Analysis and Understanding [article]

Ahmed Nabil Mohamed
2015 arXiv   pre-print
Sixth, we examine many of the most cited and recent reviews in the field that have been published during the past two decades to reveal various approaches used for implementing different stages of the  ...  Moreover, we provide a long list of public datasets and discuss briefly some examples of these datasets. Finally, we provide a general discussion of the subject from the aspect of computer vision.  ...  Aggarwal and Park [9] categorized processes involved in human activities understanding into two levels: low-level vision processes such as segmentation, tracking, pose recovery, and trajectory estimation  ... 
arXiv:1509.01074v1 fatcat:kpqf277iw5f2bmlmog2ya2awci

Where's my mesh? An exploratory study on model-free grasp planning

FABRIZIO BOTTAREL
2021
In such cases, approaches that allow planning for optimal grasps given partial 3D information are essential in order to attempt manipulation actions, and the research field is far from being depleted.  ...  According to this line of thought, we propose a grasp planning method that constrains the superquadric parameterization to account for the characteristics of the target scenario.  ...  Grasp planning is a field with more than three decades worth of literature and contains a vast landscape of approaches, each with varying assumptions and methodology [50, 100, 10,  ... 
doi:10.15167/bottarel-fabrizio_phd2021-05-12 fatcat:n27hhkxhdvbuvlwuhuzizoa23e
« Previous Showing results 1 — 15 out of 22 results