A Bayesian Mixture Model for Multi-View Face Alignment

Zhou, Yi; Zhang, Wei; Tang, Xiaoou; Shum, Harry

doi:10.1109/cvpr.2005.17

Cited by 28 publications

(31 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although multi-view face shape models [6,34,35] partially solve the pose variation problem, they cannot cover unlimited possibilities of view changes. Therefore, 3D shape model [15,29] is proposed to handle continuous view change.…”

Section: Introductionmentioning

confidence: 99%

Pose-Free Facial Landmark Fitting via Optimized Part Mixtures and Cascaded Deformable Shape Model

Huang

Zhang

et al. 2013

2013 IEEE International Conference on Computer Vision

196

187

View full text Add to dashboard Cite

show abstract

Section: Introductionmentioning

confidence: 99%

Pose-Free Facial Landmark Fitting via Optimized Part Mixtures and Cascaded Deformable Shape Model

Huang

Zhang

et al. 2013

2013 IEEE International Conference on Computer Vision

196

187

View full text Add to dashboard Cite

show abstract

“…There have been some previous works extending ASM to multi-view cases, such as [6,11]. In this paper, we mainly focus on the parameter estimation approach.…”

Section: Multi-view Asm Frameworkmentioning

confidence: 99%

“…By learning statistical distributions of shapes and textures from training data, the method can be used to localize objects such as human faces [4,5,6,9,10,11,12].…”

Section: Introductionmentioning

confidence: 99%

“…These works often involve multiple shape and texture models built by non-linear methods such as Gaussian mixtures [1] and Kernel PCA [6], while self-occlusion can also be handled by learning visibility from training data [11]. However, since the texture model used in the local search of each label point depends on the view, these methods are often very sensitive to the estimation of the hidden view parameter.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Multi-View Active Shape Model with Robust Parameter Estimation

Zhang

2006

18th International Conference on Pattern Recognition (ICPR'06)

View full text Add to dashboard Cite

Active Shape Model is an efficient way for localizing objects with variable shapes. When ASM is extended to multiview cases, the parameter estimation approaches in previous works are often sensitive to the initial view, as they do not handle the unreliability of local texture search, which can be caused by bad initialization or cluttered background. To overcome this problem, we propose a novel algorithm for parameter estimation, using robust estimators to remove outliers. By weighting dynamically, our method acts as a model selection method, which reveals the hidden shape and view parameters from noisy observations of local texture models. Experiments and comparisons on multi-view face alignment are carried out to show the efficiency of our approach.

show abstract

“…In the literature of MVFA, non-linear modeling method such as Gaussian Mixture Model [6], kernel PCA [7], Bayesian Mixture Model with learning visibility of label points [8] and view-based methods such as view based DAM [9] and view-based ASM [10] are developed which are mainly 2D approaches with no appealing to 3D face information. Due to the intrinsic difficulties caused by face appearance changes in 2D face images of a 3D face, MVFA is still not a solved problem.…”

Section: Introductionmentioning

confidence: 99%

Multi-View Face Alignment Using 3D Shape Model for View Estimation

Lao

2009

Advances in Biometrics

View full text Add to dashboard Cite

Abstract. For multi-view face alignment (MVFA), the non-linear variation of shape and texture, and the self-occlusion of facial feature points caused by view change are the two major difficulties. The state-of-the-art MVFA methods are essentially view-based approaches in which views are divided into several categories such as frontal, half profile, full profile etc. and each of them has its own model in MVFA. Therefore the view estimation problem becomes a critical step in MVFA. In this paper, a MVFA method using 3D face shape model for view estimation is presented in which the 3D shape model is used to estimate the pose of the face thereby selecting its model and indicating its selfoccluded points. Experiments on different datasets are reported to show the improvement over previous works. [10] are developed which are mainly 2D approaches with no appealing to 3D face information. Due to the intrinsic difficulties caused by face appearance changes in 2D face images of a 3D face, MVFA is still not a solved problem. KeywordsThe state-of-the-art MVFA methods are essentially view-based approaches in which views are divided into several categories such as frontal, half profile, full profile etc. and each of them has its own shape and texture models. Since the texture model used in local search of each label point of a particular shape model depends on

show abstract

A Bayesian Mixture Model for Multi-View Face Alignment

Cited by 28 publications

References 8 publications

Pose-Free Facial Landmark Fitting via Optimized Part Mixtures and Cascaded Deformable Shape Model

Pose-Free Facial Landmark Fitting via Optimized Part Mixtures and Cascaded Deformable Shape Model

Multi-View Active Shape Model with Robust Parameter Estimation

Multi-View Face Alignment Using 3D Shape Model for View Estimation

Contact Info

Product

Resources

About