2,897 Hits in 4.1 sec

Efficient constrained local model fitting for non-rigid face alignment

Simon Lucey, Yang Wang, Mark Cox, Sridha Sridharan, Jeffery F. Cohn
2009 Image and Vision Computing  
Active appearance models (AAMs) have demonstrated great utility when being employed for non-rigid face alignment/tracking.  ...  In this paper we introduce an extension to a discriminative method for non-rigid face registration/tracking referred to as a constrained local model (CLM).  ...  Input:-template ðTÞ, source image ðYÞ, steepest descent matrix (K) initial warp guess (p), index to the template (z), threshold ðÞ Output:-final warp (p) (1) Warp the source image Y with z 0 ¼ Wðz; pÞ.  ... 
doi:10.1016/j.imavis.2009.03.002 pmid:20046797 pmcid:PMC2799037 fatcat:ped6z77hgnffxk6ibx66mb4rcu

Simultaneous Reconstruction and Tracking of Non-planar Templates [chapter]

Sebastian Lieberknecht, Selim Benhimane, Slobodan Ilic
2011 Lecture Notes in Computer Science  
In this paper, we address the problem of simultaneous tracking and reconstruction of non-planar templates in real-time.  ...  Classical approaches to template tracking assume planarity and do not attempt to recover the shape of an object.  ...  Silveira and Malis [17] use 2D warps and present a generic framework for template tracking which can undergo deformations.  ... 
doi:10.1007/978-3-642-23123-0_40 fatcat:y22z6tw5q5ho3p6nvzehwnsyiq

Facial tracking in video transfer using watershed segmentation

Mayilvahanan P, Purushothaman S, Rajeswari P
2014 International Journal of Engineering & Technology  
This paper presents the implementation of watershed segmentation for facial tracking. It focuses on tracking face of human in a continuous video transfer.  ...  Tracking efficiency is 97.2%.  ...  The output of the face tracker can be the 2D position of the face in each image of the video (2D tracking), the 3D pose of the face (3D tracking), or the location of features on the face.  ... 
doi:10.14419/ijet.v3i4.3633 fatcat:dnxvb2u7x5dajd7at5iol7zrye

Fast, reliable head tracking under varying illumination: an approach based on registration of texture-mapped 3D models

M. La Cascia, S. Sclaroff, V. Athitsos
2000 IEEE Transactions on Pattern Analysis and Machine Intelligence  
Tracking does not require a precise initial fit of the model; the system is initialized automatically using a simple 2D face detector.  ...  The regularization term tends to limit potential ambiguities that arise in the warping and illumination templates. It enables stable tracking over extended sequences.  ...  In another approach [10] , flow is used to constrain the motion of an anatomically-motivated face model and integrated with edge forces to improve tracking results.  ... 
doi:10.1109/34.845375 fatcat:ltwzadlbgjaqpdvyfueo7h4ppy

The template update problem

L. Matthews, T. Ishikawa, S. Baker
2004 IEEE Transactions on Pattern Analysis and Machine Intelligence  
Template tracking dates back to the 1981 Lucas-Kanade algorithm.  ...  One question that has received very little attention, however, is how to update the template so that it remains a good model of the tracked object.  ...  For example, tracking with an affine warp is often performed by first tracking with a translation, then a 2D similarity transformation, and finally a full affine warp.  ... 
doi:10.1109/tpami.2004.16 pmid:18579941 fatcat:uf7agvkjmrccfkt42nwjc6yucu

The Template Update Problem

I. Matthews, T. Ishikawa, S. Baker
2003 Procedings of the British Machine Vision Conference 2003  
Template tracking dates back to the 1981 Lucas-Kanade algorithm.  ...  One question that has received very little attention, however, is how to update the template so that it remains a good model of the tracked object.  ...  For example, tracking with an affine warp is often performed by first tracking with a translation, then a 2D similarity transformation, and finally a full affine warp.  ... 
doi:10.5244/c.17.65 dblp:conf/bmvc/MatthewsIB03 fatcat:l6inyv34jngmzfh7heoq2453ue

Multimedia Databases for Video Indexing: Toward Automatic Face Image Registration

Simon Clippingdale, Mahito Fujii, Masahiro Shibata
2009 2009 11th IEEE International Symposium on Multimedia  
In this paper we outline an algorithm for estimating the generalized pose and 3D feature point locations from 2D feature point tracking data that may be noisy and subject to occlusions, and a texture mapping  ...  We aim to automate the registration process by constructing 3D face models from the imagery available for registration, and then using the constructed models to generate templates for the face recognition  ...  The face tracking and recognition system uses 2D deformable templates, and handles 3D head rotation by registering templates of each target individual at multiple head rotations (poses).  ... 
doi:10.1109/ism.2009.91 dblp:conf/ism/ClippingdaleFS09 fatcat:zsbypx2i3jeobhegbqaxd2itjm

Video Face Editing Using Temporal-Spatial-Smooth Warping [article]

Xiaoyan Li, Dacheng Tao
2014 arXiv   pre-print
In particular, TSSW is robust to subtly inaccurate localization of feature points and is a vast improvement over image-based warping methods.  ...  Simply applying image-based warping algorithms to video-based face editing produces temporal incoherence in the synthesized videos because it is impossible to consistently localize facial features in two  ...  the help of a training dataset; 2) for transferring makeup, a similar template face to be warped (with similar skin color to that of the input face) is selected from a template dataset.  ... 
arXiv:1408.2380v1 fatcat:yb5foxdosnfmzdo66ggwtxf4dm

Unconstrained 3D face reconstruction

Joseph Roth, Yiying Tong, Xiaoming Liu
2015 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
2D images is a long-standing computer vision problem.  ...  The output of our algorithm is a true 3D face surface model represented as a watertight triangulated surface with albedo data or texture information. 3D face reconstruction from a collection of unconstrained  ...  Based on this observation, we aim to warp the template in a way such that the projections of the warped 3D landmark locations can match well with the estimated 2D landmarks.  ... 
doi:10.1109/cvpr.2015.7298876 dblp:conf/cvpr/RothTL15 fatcat:viiemsm7yjeqdab5cchulyaxte

Mask-off: Synthesizing Face Images in the Presence of Head-mounted Displays [article]

Yajie Zhao, Qingguo Xu, Xinyu Huang, Ruigang Yang
2016 arXiv   pre-print
In face synthesis, we propose a novel algorithm that can robustly align and track a personalized 3D head model given a face that is severely occluded by the HMD.  ...  To this end, we propose a novel synthesis framework that contains four modules: 3D head reconstruction, face alignment and tracking, face synthesis, and eye synthesis.  ...  Acknowledgements Thank Chen Cao and Kun Zhou to share their face datasets.  ... 
arXiv:1610.08481v2 fatcat:v2cd7kuee5hzrg3z2gtvz4l37y

Face Flow

Patrick Snape, Anastasios Roussos, Yannis Panagakis, Stefanos Zafeiriou
2015 2015 IEEE International Conference on Computer Vision (ICCV)  
As experimental evaluation, we show quantitative experiments on a challenging novel benchmark of face sequences, with dense ground truth optical flow provided by motion capture data.  ...  We formulate a novel energy minimisation problem for establishing dense correspondences between a neutral template and every frame of a sequence.  ...  Stefanos Zafeiriou is partially supported by the EPSRC project EP/J017787/1 (4D-FAB) and is also partially supported by the EPSRC project EP/L026813/1 Adaptive Facial Deformable Models for Tracking (ADAManT  ... 
doi:10.1109/iccv.2015.343 dblp:conf/iccv/SnapeRPZ15 fatcat:yisgb3ihuzg4dporobos3ojuey

Monocular and Stereo Methods for AAM Learning from Video

Jason Saragih, Roland Goecke
2007 2007 IEEE Conference on Computer Vision and Pattern Recognition  
First, we propose a new method for automatic correspondence finding in monocular images which is based on an adaptive template tracking paradigm.  ...  The performance of the monocular method was first evaluated on a pre-annotated database of a talking face.  ...  For example, the Jacobian of the data term in Equation (13) is now given by: J D (x) = −∇I( x) ∂ x ∂p 2D ∂p 2D ∂p 3D , T(x), 1 , (39) where ∂p 2D ∂p 3D is the derivative of the 2D landmarks with respect  ... 
doi:10.1109/cvpr.2007.383058 dblp:conf/cvpr/SaragihG07 fatcat:afjb7ppfufh55bfikxedu5jw4a

Multiscale Inverse Compositional Alignment for Subdivision Surface Maps [chapter]

Igor Guskov
2004 Lecture Notes in Computer Science  
We propose an efficient alignment method for textured Doo-Sabin subdivision surface templates.  ...  The method is applied to real-time tracking of specially marked surfaces from a single camera view.  ...  This improves the convergence properties of the matching without resorting to eigen-tracking approaches.  ... 
doi:10.1007/978-3-540-24670-1_11 fatcat:scsf5oowzzbm5fwuejl3qnaiyy

AAM based HCI for an intelligent wheelchair

Pei Jia, Huosheng Hu, Yongji Wang, Jun Li, Bangjun Lei, Jingyu Yang
2007 MIPPR 2007: Remote Sensing and GIS Data Processing and Applications; and Innovative Multispectral Technology and Applications  
In every tracking frame, AAM fitting is iterated until the current face is converged to the template trained face.  ...  In this paper, the 2D inverse compositional image alignment algorithm is adopted as our face tracking module.  ...  Since the user's face may not be quite upright sometimes, AAM can also be used to deal with this face rotation situation to some extend so that the face turning angles can be correctly tracked based on  ... 
doi:10.1117/12.749184 fatcat:adufidhlwzd2lesbtx2c5mfnki

Non-Rigid Object Alignment with a Mismatch Template Based on Exhaustive Local Search

Yang Wang, Simon Lucey, Jeffrey Cohn
2007 2007 IEEE 11th International Conference on Computer Vision  
Specifically, we fit the patch-based face model to an unseen image using an exhaustive local search and constrain the local warp updates within a global warping space.  ...  Non-rigid object alignment is especially challenging when only a single appearance template is available and target and template images fail to match.  ...  The size of the template face region is 110 × 110 pixels with the inter-ocular distance of 45 pixels.  ... 
doi:10.1109/iccv.2007.4409188 dblp:conf/iccv/WangLC07 fatcat:x7zqzerckzfe3jfl75675h6cla
« Previous Showing results 1 — 15 out of 2,897 results