G. W. Cottrell and J. Metcalfe. EMPATH: Face, Gender and Emotion Recognition using Holons. In R. P. Lippman, J. Moody, and D. S. Touretzky, editors, Advances in Neural Information Processing Systems, volume 3, pages 564-571, 1991.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... variations occur due to scale changes as well as in plane and out of plane rotations Deformation Extraction Holistic Methods Local Methods Image based Neural network [18] 35] 54] Intensity pro les [2] Gabor wavelets [18] 35] High gradient components [52] PCA Neural networks [66][17] Model based Active appearance model [51] 16] 24] Geometric face model [49] Point distribution model [41] 2 view point based models [68] Labeled graphs [86] 56] 40] Motion Extraction Holistic Methods Local Methods Dense Optical Flow Dense ow elds [52] 2] Region based ow ....

....face processing also with lighting variations. Common for most holistic face analysis approaches is the need of a thorough face background separation in order to prevent disturbance caused by clutter. Local Image based Approaches: Padgett and Cottrell [66] as well as Cottrell and Metcalfe [17] extracted facial expressions from windows placed around intransient facial feature regions (both eyes and mouth) and employed local Principal Component Analysis (PCA) for representation purposes. Local transient facial features such as wrinkles can be measured by using image intensity pro les ....

G. W. Cottrell and J. Metcalfe. EMPATH: Face, Gender and Emotion Recognition using Holons. In R. P. Lippman, J. Moody, and D. S. Touretzky, editors, Advances in Neural Information Processing Systems, volume 3, pages 564-571, 1991.

....PCA can only separate pairwise linear dependencies between pixels. High order dependencies will still show in the joint distribution of PCA coecients, and thus will not be properly separated. Some of the most successful representations for face recognition, such as eigenfaces [57] holons [15], and local feature analysis [50] are based on PCA. In a task such as face recognition, much of the important information may be contained in the high order relationships among the image pixels and thus it is important to investigate whether generalizations of PCA which are sensitive to ....

G. Cottrell and J. Metcalfe. Face, gender and emotion recognition using holons. In D. Touretzky, editor, Advances in Neural Information Processing Systems, volume 3, pages 564-571, San Mateo, CA, 1991. Morgan Kaufmann.

....facial expression recognition. These systems include measurement of facial motion through optic flow [16, 23, 21, 9] measurements of the shapes of facial features and their spatial arrangements [13] holistic spatial pattern analysis using techniques based on principal components analysis (PCA) [2, 19, 13] and methods for relating face images to physical models of the facial skin and musculature [16, 22, 14, 9] Most of the previous work employed datasets of posed expressions collected under controlled image conditions. Subjects deliberately faced the camera and the facial expressions were ....

G. Cottrell and J. Metcalfe. Face, gender and emotion recognition using holons. In D. Touretzky, editor, Advances in Neural Information Processing Systems, volume 3, pages 564--571, San Mateo, CA, 1991. Morgan Kaufmann.

....expressions from static images. Turk and Pentland [20] represent face images by eigenfaces through linear principal component analysis. Padgett and Cottrell [14] use an approach similar to eigenfaces but with seven pixel blocks from feature regions (both eyes and mouth) Cottrell and Metcalfe [5] use holistic representations based on principal components, extracted by feed forward networks. Rahardja et al. 15] also use holistic representations with neural networks, but the images are represented in a pyramid structure. Lanitis et al. 10] use parameterized deformable templates (flexible ....

G. Cottrell and J. Metcalfe. Face, gender and emotion recognition using holons. In D. Touretzky, editor, Advances in Neural Information Processing Systems 3, pages 564--571. Morgan and Kaufman, San Mateo, 1991.

....kernels. Recognition across changes in pose with the ICA representation was 93 , compared to 87 with a PCA representation, and across changes in lighting ICA gave 100 correct recognition, compared to 90 with PCA. Introduction Important advances in face recognition such as Holons (Cottrell Metcalfe, 1991) and Eigenfaces (Turk Pentland 1991) have employed forms of principal component analysis, which addresses only second order moments of the input (Cottrell Metcalfe, 1991; Turk Pentland 1991) Independent component analysis (ICA) is a generalization of principal component analysis (PCA) ....

....correct recognition, compared to 90 with PCA. Introduction Important advances in face recognition such as Holons (Cottrell Metcalfe, 1991) and Eigenfaces (Turk Pentland 1991) have employed forms of principal component analysis, which addresses only second order moments of the input (Cottrell Metcalfe, 1991; Turk Pentland 1991) Independent component analysis (ICA) is a generalization of principal component analysis (PCA) which decorrelates the higher order moments of the input (Comon, 1994) In a task such as face recognition, much of the important information is contained in the high order ....

Cottrell & Metcalfe, 1991. Face, gender and emotion recognition using Holons. In Advances in Neural Information Processing Systems 3, D. Touretzky, (Ed.), Morgan Kaufman, San Mateo, CA: 564 - 571.

....a survey and comparison of recent techniques for facial expression recognition as applied to automated FACS encoding. Recent approaches include measurement of facial motion through optic flow [44, 64, 54, 26, 15, 43] and analysis of surface textures based on principal component analysis (PCA) [17, 48, 40]. In addition, a number of methods that have been developed for representing faces for identity recognition may also be powerful for expression analysis. These approaches are also included in the present comparison. These include Gabor wavelets [20, 39] linear discriminant analysis [8] local ....

....facial expressions in images. The approaches that have been explored include analysis of facial motion [44, 64, 54, 26] measurements of the shapes of facial features and their spatial arrangements [40, 66] holistic spatial pattern analysis using techniques based on principal component analysis [17, 48, 40] graylevel pattern analysis using local spatial filters [48, 66] and methods for relating face images to physical models of the facial skin and musculature [44] 59, 42, 26] The image analysis techniques in these systems are relevant to the present goals, but the systems themselves are of limited ....

[Article contains additional citation context not shown here]

G. Cottrell and J. 1991 Metcalfe. Face, gender and emotion recognition using holons. In D. Touretzky, editor, Advances in Neural Information Processing Systems, volume 3, pages 564--571, San Mateo, CA, 1991. Morgan Kaufmann.

....6 AU 7 Figure 2: Examples of the six actions used in this study. AU 1: Inner brow raiser. 2: Outer brow raiser. 4: Brow lower. 5: Upper lid raiser (widening the eyes) 6: Cheek raiser. 7: Lid tightener (partial squint) 3 HOLISTIC SPATIAL ANALYSIS The Eigenface (Turk Pentland, 1991) and Holon (Cottrell Metcalfe, 1991) representations are holistic representations based on principal components, which can be extracted by feed forward networks trained by back propagation. Previous work in our lab and others has demonstrated that feed forward networks taking such holistic representations as input can successfully ....

....based on principal components, which can be extracted by feed forward networks trained by back propagation. Previous work in our lab and others has demonstrated that feed forward networks taking such holistic representations as input can successfully classify gender from facial images (Cottrell Metcalfe, 1991; Golomb, Lawrence, Sejnowski, 1991) We evaluated the ability of a back propagation network to classify facial actions given principal components of graylevel images as input. The primary difference between the present approach and the work referenced above is that we take the principal ....

Cottrell, G.,& Metcalfe, J. (1991): Face, gender and emotion recognition using holons. In Advances in Neural Information Processing Systems 3, D. Touretzky, (Ed.) San Mateo: Morgan & Kaufman. 564 - 571.

....San Diego, 1998 Terrence J. Sejnowski, Dissertation Adviser Donald I. A. Macleod, Committee Chair In a task such as face recognition, much of the important information may be contained in the high order relationships among the image pixels. Representations such as Eigenfaces [197] and Holons [48] are based on Principal component analysis (PCA) which encodes the correlational structure of the input, but does not address high order statistical dependencies such as relationships among three or more pixels. Independent component analysis (ICA) is a generalization of PCA which encodes the ....

....such decorrelation mechanisms as a general coding strategy for the visual system. Some of the most successful algorithms for face recognition are based on learning mechanisms that are sensitive to the correlations in the face images. For example, representations such as Eigenfaces [197] Holons [48] and Local Feature Analysis [156] are data driven face representations based on principal component analysis. Principal component analysis separates the correlations in the input, but does not address high order dependencies such as the relationships among three or more pixels. Edges are an ....

[Article contains additional citation context not shown here]

G. Cottrell and J. . Metcalfe. Face, gender and emotion recognition using holons. In D. Touretzky, editor, Advances in Neural Information Processing Systems, volume 3, pages 564--571, San Mateo, CA, 1991. Morgan Kaufmann.

....expressions from static images. Turk and Pentland [20] represent face images by eigenfaces through linear principal component analysis. Padgett and Cottrell [14] use an approach similar to eigenfaces but with seven pixel blocks from feature regions (both eyes and mouth) Cottrell and Metcalfe [5] use holistic representations based on principal components, extracted by feed forward networks. Rahardja et al. 15] also use holistic representations with neural networks, but the images are represented in a pyramid structure. Lanitis et al. 10] use parameterized deformable templates (flexible ....

G. Cottrell and J. Metcalfe. Face, gender and emotion recognition using holons. In D. Touretzky, editor, Advances in Neural Information Processing Systems 3, pages 564-- 571. Morgan and Kaufman, San Mateo, 1991.

....for face recognition [57] and general face processing [42, 43, 44, 58] There have been a number of studies using backpropagation networks for face processing. Among these are the works of Cottrell and Fleming [15] the Sexnet system of Golomb et al. the EMPATH system of Cottrell and Metcalfe [7], and the NLDR system of DeMers and Cottrell [13] A recent comparison of these systems [58] characterized them as three types: binary [15] linear [7] and face manifold [13] The next section discusses the wavelet transform, which is used to build a multiresolution representation of ....

....processing. Among these are the works of Cottrell and Fleming [15] the Sexnet system of Golomb et al. the EMPATH system of Cottrell and Metcalfe [7] and the NLDR system of DeMers and Cottrell [13] A recent comparison of these systems [58] characterized them as three types: binary [15] linear [7], and face manifold [13] The next section discusses the wavelet transform, which is used to build a multiresolution representation of average detail face characteristics in the first stage of the FuzzyFace system. 4 Wavelet Transform Wavelets have been used previously for face processing [12, ....

G. W. Cottrell and J. Metcalfe. EMPATH: Face, gender and emotion recognition using holons. In D. S. Touretzky and R. Lippman, editors, Advances in Neural Information Processing Systems, volume 3, pages 564--571. Kaufman, San Mateo, CA, 1991.

....or in image sequences, for face image interpretation. Techniques based on grey level information show more promise. Turk and Pentland [19] describe face identification using an eigenface expansion. The eigenface weights are used for classification. Many researchers have built on this approach [6,8], which is perhaps the most successful to date. Other approaches based on grey level information, include template matching [3] and filtering with multi scale Gabor filters [14] 3: Overview of our Approach Our approach can be divided into two main phases: modelling, in which flexible models of ....

G.W. Cottrell and J. Metcalfe. EMPATH: Face, Gender and Emotion Recognition Using Holons. Advances in Neural Information Processing Systems, Vol. 3, eds. R.P. Lippman, J. Moody and D.S. Touretzky, pp 564-571, Kauf- mann, San Mateo, 1991.

....The features generated by the hidden units are global face features that can serve as input for classification tasks. Cottrell and Fleming used the representation for face detection , recognition, and sex while Cottrell and Metcalfe applied a similar network for sex, identity and emotion [18]. Golomb et al. used this type network for their test of gender categorization [27] A related approach is to directly use the principal components of a set of training faces. Research that encodes faces as the projections onto the eigenvectors of normalized, aligned faces has been quite ....

.... a strong minority component [44] The hidden unit activations of an auto associator neural network also provide a face representation in this sub space [19] With this representation, researchers have demonstrated good identification rates [26] the ability to classify faces with respect to sex [27, 18], but only limited success in emotion recognition [18] 13 For emotion classification, the face representation for the most successful of the research carried out so far makes and achieves 80 90 recognition rates using temporal sequences of face images [39, 59] Mase s system uses optical flow ....

[Article contains additional citation context not shown here]

Garrison W. Cottrell and Janet Metcalfe. Empath: Face, gender and emotion recognition using holons. In R.P. Lippman, J. Moody, and D.S. Touretzky, editors, Advances in Neural Information Processing Systems 3, pages 564--571, San Mateo, 1991. Morgan Kaufmann.

....system was identical to the face recognition system described earlier #see Figure 4#, except that in addition to identity and gender outputs, the classi #er network had emotion labels. The resulting network, dubbed EMPATH #for EMotion PATtern recognition using Holons#, was a dismal failure #Cottrell and Metcalfe, 1991#. The Is all face processing holistic 19 system reliably identi#ed the individuals and correctly learned their gender. However, when it came to emotion labels, the system could not even learn the training set. Consistent with the observations made during image capture, the network was able to ....

Cottrell, G. W. and Metcalfe, J. #1991#. Empath: Face, gender and emotion recognition using holons. In Lippman, R. P., Moody, J., and Touretzky, D. S., editors, Advances in Neural Information Processing Systems 3, pages 564#571, San Mateo. Morgan Kaufmann.

....achieved only a 60 generalization rate. This suggests that the actual representational scheme used by the brain to identify emotions may consist of face features rather than the entire face. 1 Introduction In an extension of Cottrell and Metcalfe s work on recognizing emotions in face images [5], the performance of artificial neural networks in classification of emotions in face images is explored. In their work, undergraduates were asked to exhibit a number of different emotions. The images were then compressed with an auto associative network, and the hidden unit activations for each ....

Garrison W. Cottrell and Janet Metcalfe. Empath: Face, gender and emotion recognition using holons. In R.P. Lippman, J. Moody, and D.S. Touretzky, editors, Advances in Neural Information Processing Systems 3, pages 564--571, San Mateo, 1991. Morgan Kaufmann.

....between the two modules affected the specializations developed by the network. Figure 2: Image preprocessing: Gabor jets, separate principal components analysis on responses at each frequency. 2. 1 Image Data We acquired face images from the Cottrell and Metcalfe facial expression database [3] and captured multiple images of several books, cups, and soda cans with a CCD camera and video frame grabber. For the face images, we chose five grayscale images of each of 12 individuals. The images were photographed under controlled lighting and pose conditions; the subjects portrayed a ....

Garrison W. Cottrell and Janet Metcalfe. Empath: Face, gender and emotion recognition using holons. In Advances in Neural Information Processing Systems 3, pages 564--571, 1991.

....feign. In previous work, Cottrell and Metcalfe had undergraduates feign emotions. While their network performed well on identity and gender classification, it never did well on emotion. Cottrell and Metcalfe speculated that their results were due to poor portrayal of the emotions by their subjects [5]. Figure 2: The top 25 eigenvectors from PCA of 32x32 pixel patches drawn randomly over the face database. To reduce this possibility, we make use of a validated facial emotion database (Pictures of Facial Affect) assembled by Ekman and Friesen [6] Each of the face images in this set exhibits a ....

Garrison W. Cottrell and Janet Metcalfe. Empath: Face, gender and emotion recognition using holons. In R.P. Lippman, J. Moody, and D.S. Touretzky, editors, Advances in Neural Information Processing Systems 3, pages 564--571, San Mateo, 1991. Morgan Kaufmann.

No context found.

G. Cottrell and J. 1991 Metcalfe. Face, gender and emotion recognition using holons. In D. Touretzky, editor, Advances in Neural Information Processing Systems, volume 3, pages 564-571, San Mateo, CA, 1991. Morgan Kaufmann.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC