Active Appearance Models with Rotation Invariant Kernels

Hamsici, Onur C.; Martı́nez, Aleix M.

doi:10.1109/iccv.2009.5459365

Cited by 15 publications

(13 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Quantitative results of this pairwise optimization approach are given in Table 2, where we see that it yields more accurate results than other state of the art algorithms (21,22). In fact, these results are quite close to the detection errors obtained with manual annotations, which are known to be between 4.1 and 5.1 pixels in images of this complexity (18).…”

Section: Significancesupporting

confidence: 59%

Compound facial expressions of emotion

Tao

Martı́nez

2014

Proc. Natl. Acad. Sci. U.S.A.

632

467

View full text Add to dashboard Cite

Understanding the different categories of facial expressions of emotion regularly used by us is essential to gain insights into human cognition and affect as well as for the design of computational models and perceptual interfaces. Past research on facial expressions of emotion has focused on the study of six basic categories-happiness, surprise, anger, sadness, fear, and disgust. However, many more facial expressions of emotion exist and are used regularly by humans. This paper describes an important group of expressions, which we call compound emotion categories. Compound emotions are those that can be constructed by combining basic component categories to create new ones. For instance, happily surprised and angrily surprised are two distinct compound emotion categories. The present work defines 21 distinct emotion categories. Sample images of their facial expressions were collected from 230 human subjects. A Facial Action Coding System analysis shows the production of these 21 categories is different but consistent with the subordinate categories they represent (e.g., a happily surprised expression combines muscle movements observed in happiness and surprised). We show that these differences are sufficient to distinguish between the 21 defined categories. We then use a computational model of face perception to demonstrate that most of these categories are also visually discriminable from one another.

show abstract

Section: Significancesupporting

confidence: 59%

Compound facial expressions of emotion

Tao

Martı́nez

2014

Proc. Natl. Acad. Sci. U.S.A.

632

467

View full text Add to dashboard Cite

show abstract

“…The algorithms consisted of KRR or ε -SVR with pixel intensities or C1 features, the nonlinear AAM of [9], and Adaboost regression with Haar fetures as in SRM [13]. The goals were to evaluate how well each algorithm: generalizes across several identities, manages occlusions and extreme expression changes, handles extreme degradation in resolution, and performs in a real world setting.…”

Section: Methodsmentioning

confidence: 99%

“…This method was enhanced by using boosting to learn the shape parameter update and confidence score [6]. Other authors take a Bayesian approach to shape modeling, learning the conditional density of the shape parameters given the object image, and iterating to the maximum a posteriori (MAP) estimate of the shape parameters [7, 8, 9]. Liang et al [10] improve on the idea by using regularization at accurately aligned points to reduce the occurence of local minima favored by the global shape model.…”

Section: Introductionmentioning

confidence: 99%

Learning deformable shape manifolds

Rivera

Martı́nez

2012

Pattern Recognition

View full text Add to dashboard Cite

We propose an approach to shape detection of highly deformable shapes in images via manifold learning with regression. Our method does not require shape key points be defined at high contrast image regions, nor do we need an initial estimate of the shape. We only require sufficient representative training data and a rough initial estimate of the object position and scale. We demonstrate the method for face shape learning, and provide a comparison to nonlinear Active Appearance Model. Our method is extremely accurate, to nearly pixel precision and is capable of accurately detecting the shape of faces undergoing extreme expression changes. The technique is robust to occlusions such as glasses and gives reasonable results for extremely degraded image resolutions.

show abstract

“…Xiao et al [33] also develop the research work of combining 2D AAM and 3D Morphable Model (3DMM). Hamsici and Martinez [16] derive a new approach carries the advantages of AAM and 3DMM that can model nonlinear changes in examples without the need of a pre-alignment step. Lee and Kim [21] propose a tensor-based AAM that can handle a variety of subjects, poses, expressions, and illuminations in the tensor algebra framework.…”

mentioning

confidence: 99%

Facial Landmark Localization

Ding

Wang

2011

Handbook of Face Recognition

View full text Add to dashboard Cite

Face detection and recognition is a vibrant area of biometrics with active research and commercial efforts over the last 20 years. The task of face detection is to search faces in images, reporting their positions by a bounding box. Recent studies [19,31] have shown that face detection has already been a state-of-the-art technology in both accuracy and speed. However, face detection is not sufficient to acquire facial landmarks, for example, eye contours, mouth corners, nose, eyebrows, etc. This is therefore the task of facial landmark localization which aims to find the accurate positions of the facial feature points as illustrated in Fig. 12.1. It is a fundamental and significant work in face-related areas, for example, face recognition, face cartoon/sketch, face pose estimate, model-based face tracking, eye/mouth motion analysis, 3D face reconstruction, etc.There is a wide variety of works related to facial landmark localization. The early researches extract facial landmarks without a global model. Facial landmarks, such as the eye corners and centers, the mouth corners and center, the nose corners, chin and cheek borders are located based on geometrical knowledge. The first step consists of the establishment of a rectangular search region for the mouth and a rectangular search region for the eyes. The borders are extracted by applying corner detection algorithm such as SUSAN border extraction algorithm [17]. Such methods are fast, however, they could not deal with faces of large variation in appearance due to pose, rotation, illumination and background changes.X. Ding ( ) · L. Wang

show abstract

Active Appearance Models with Rotation Invariant Kernels

Abstract: 2D Active Appearance Models (AAM) and 3D Morphable Models (3DMM)

Cited by 15 publications

References 14 publications

Compound facial expressions of emotion

Compound facial expressions of emotion

Learning deformable shape manifolds

Facial Landmark Localization

Contact Info

Product

Resources

About