A Common Framework for Real-Time Emotion Recognition and Facial Action Unit Detection

Size: px
Start display at page:

Download "A Common Framework for Real-Time Emotion Recognition and Facial Action Unit Detection"

Transcription

1 A Common Framework for Real-Time Emotion Recognition and Facial Action Unit Detection Tobias Gehrig and Hazım Kemal Ekenel Facial Image Processing and Analysis Group, Institute for Anthropomatics Karlsruhe Institute of Technology D Karlsruhe, P.O. Box 6980 Germany {tobias.gehrig, Abstract In this paper, we present a common framework for realtime action unit detection and emotion recognition that we have developed for the emotion recognition and action unit detection sub-challenges of the FG 2011 Facial Expression Recognition and Analysis Challenge. For these tasks we employed a local appearance-based face representation approach using discrete cosine transform, which has been shown to be very effective and robust for face recognition. Using these features, we trained multiple one-versusall support vector machine classifiers corresponding to the individual classes of the specific task. With this framework we achieve 24.2% and 7.6% absolute improvement over the overall baseline results on the emotion recognition and action unit detection sub-challenge, respectively. 1. Introduction Facial expressions are naturally used by humans as one way to communicate their emotions, opinions, intentions, and cognitive states with others and are therefore important in natural communication. However, in today s humancomputer interaction (HCI) scenarios this kind of information is mostly being neglected, in spite of the fact that it can be a much faster and more direct way than describing the affective state with words. To compensate for this neglect and improve HCI, recently, there have been many studies conducted on automatic facial expression analysis and the topic has become more and more popular. A wide range of applications [5] including, but not limited to, psychological studies, pain [2] or stress detection, online tutoring systems [19], and assistance systems for autistic persons [12] has also fueled this increasing interest in this topic. When we focus on applications in HCI scenarios, such as mobile service robots [20], or especially safety critical applications, such as drowsy driver detection [18], the runtime of an algorithm plays an important role, since the system should react in real-time without much latency. Facial expression analysis can be performed in different degrees of granularity. It can be either done by directly classifying the prototypic expressions (e.g. anger, fear, joy, surprise,... ) from face images or, with finer granularity, by detecting facial muscle activities. The latter is commonly described using the facial action coding system (FACS) [8], which defines action units (AU) corresponding to atomic facial muscle actions. Previous approaches to automatic expression analysis utilized, besides other feature representations, gabor wavelets [4], local binary patterns (LBP) [15] or the widely used active appearance models (AAM) to model the human face [11, 14]. Detailed surveys about works on automatic expressions analysis can be found in [9, 13, 16, 21]. In this study, we considered two main points while choosing our face representation method. The first main point was to be able to perform multiple face classification tasks, such as face recognition, gender and expression classification, with the same representation. The reason behind this is that if we can have a common framework that can provide us, for example, identity and gender information of the subject, this information can then be used to enhance the performance of the expression analysis system by employing person specific or gender specific expression analysis models, which has been shown in [14] to improve the performance. The second main point was to be able to have real-time processing capability. To achieve this, the feature extraction should be computed fast and at the same time the representation should be compact. These design choices led us to utilize the discrete cosine transform (DCT) for the representation. We employed a local appearancebased face representation approach, in which DCT is used to model local facial regions [7], and an ensemble of support vector machine (SVM) classifiers for the task of automatic facial expression analysis. We evaluate the approach 1

2 MCT-based face & eye detection eye-based alignment block-based DCT 10 Coeff. per block 1-vs-all SVM (RBF kernel) Figure 1. Overview of the common face processing and classification framework. within the emotion recognition and action unit detection sub-challenges of the FG 2011 Facial Expression Recognition and Analysis Challenge (FERA2011) [17] and compare our results to the baseline results, which were obtained using a local binary pattern (LBP)-based face representation [17]. We achieve 24.2% absolute improvement in terms of classification rate over the baseline results for the emotion recognition sub-challenge, and 7.6% absolute improvement in terms of F1-score for the sub-challenge on action unit detection using the same face representation. These results lead to a high rank and outperformed a range of other approaches in the FERA2011 emotion recognition and action unit detection sub-challenges [1]. In Section 2, we will describe the face representation framework, followed by a description of FERA2011, its dataset, tasks, and baseline system in Section 3. Then, we will present our proposed system for both tasks and discuss their results in Section 4. Finally, in Section 5, we will give conclusions and future research directions. 2. Methodology The face processing system is illustrated in Figure 1 and described in more detail in the following subsections Preprocessing The face and eyes are automatically detected using a modified census transformation (MCT) based face and eye detector [10]. Then the face image is aligned by an euclidean transform using the detected eye coordinates so that for all images the eyes are at the same position based on a given eye row and an interocular distance. This reduces the amount of variation in the feature space that is due to in-plane rotation, scale variations, and small pose changes. Finally, the image is converted to grayscale Local Appearance-based Face Representation From the preprocessed face image a local appearancebased face representation is retrieved. Such representations are known to provide more robustness against local appearance changes than holistic ones. This is due to the fact that the face representation of local approaches differs only in those regions, where changes occur, e.g. due to expression, local illumination, or occlusion, whereas for holistic approaches the whole face representation changes. Therefore, they are widely used in automatic face analysis, like face recognition [7]. Here, we retrieve these local regions by dividing the face image into equally sized, nonoverlapping blocks of N N pixels, as shown in Figure 1. The block size is chosen such, that it provides stationarity, simple transform complexity, and sufficient compression. These local blocks are then processed individually by the two-dimensional type-ii DCT, which provides a compact representation of the data similar to the Karhunen-Loeve transform (KLT), but with the advantage of being data independent. The DCT coefficients are extracted using zig-zag scanning and all coefficients but the first few are removed. To reduce the effect of illumination and balance the contribution of each coefficient, which normally has higher magnitude for small indices, the resulting coefficient vectors are normalized separately for each block, as proposed by Ekenel [7]. This is achieved by first dividing each coefficient by its standard deviation and normalizing the resulting local feature vector to unit norm. Finally, all local feature vectors are concatenated to form the overall feature vector Support Vector Machines (SVM) One of the most popular approaches of doing facial expression analysis is using SVMs [4, 11, 15, 17]. To prevent some attributes from dominating others, due to different numeric ranges, we normalize the feature vectors F = {f i }, before feeding them into the SVM, by making each attribute f i,j zero-mean and unit-variance over all feature vectors F. The normalization parameters are estimated on the training set and applied to the test set before classification. Here, we decided to use a SVM with a radial basis function (RBF) kernel, as proposed by [15], which transforms the features into a higher dimensional space, where it can be then linearly separated by a hyperplane. For both sub-challenges we utilize a single one-versusall classifier per class to be detected, which is trained on the occurances of the class as positive samples and occurances of samples of all the other classes as negative samples. For action unit detection, a threshold on the distance to the hyperplane serves us directly with the detection estimate for the particular AU in the current frame. For emotion recognition, additionally a model for probability estimates is trained for each classifier to make the classifier outputs more comparable. The decision for a frame is then achieved by selecting the class with the highest probability. Finally, majority voting over all frame decisions of a video gives us the emotion estimate for that video. 2

3 3. FG 2011 Facial Expression Recognition and Analysis Challenge (FERA2011) The presented approach is evaluated within the FG 2011 Facial Expression Recognition and Analysis Challenge (FERA2011) [17]. The aim of the challenge is to overcome the lack of standardized evaluation procedures and thus the low comparability in the previous literature on automatic facial expression analysis. In this section, we describe briefly the database used in FERA2011, its subchallenges, and the official baseline system. Detailed informations can be found in the challenge paper [17] Database The GEMEP-FERA database used for the FERA challenge is a subset derived from the GEneva Multimodal Emotion Portrayals (GEMEP) database [3]. It consists of recordings of 10 actors displaying a variety of expressions, while either uttering a meaningless phrase or the word Aaah. The movement of the actors faces varies from steady frontal to moving fast and wild with out-of-plane rotations. The sequences are between 1 and 4 seconds long. For the evaluation the dataset is provided as a strictly divided training and test set. The training set contains recordings of 7 actors, of which 3 are also present in the test set together with 3 unseen persons. Thus, the system can be evaluated on subject-dependent and subject-independent data Emotion Recognition Sub-Challenge For the emotion recognition sub-challenge the task is to predict one of five discrete, mutually-exclusive emotion classes (anger, fear, joy, relief, and sadness) per video. The performance of each emotion classifier is measured by means of the classification rate, meaning the ratio between the number of correctly classified videos and the total number of videos of the corresponding emotion. The overall system performance is computed as the average over all individual emotion classification rates Action Unit Detection Sub-Challenge For the action unit (AU) detection sub-challenge the task is to detect the presence or absence of 12 AUs from the facial action coding system (FACS) (1, 2, 4, 6, 7, 10, 12, 15, 17, 18, 25, and 26) on a frame-by-frame basis. During speech, which was labeled as AD50, AU25 and AU26 were not labeled and thus these speech frames are neither used for the training of the AU25 and AU26 classifiers nor are they included in the computation of the scores. The performance of each AU detector is measured by means of the F1- measure and the overall system performance is computed as the average over all individual F1-scores Baseline system The organizers of FERA2011 provide baseline results on the GEMEP-FERA corpus to have a common ground for participants to compare their results to [17]. This baseline system first uses the output of the OpenCV face and eye detectors to align and scale the face image to pixels. The aligned face image is then split into blocks of pixels, for which histograms of uniform local binary patterns (ULBP) with 8 neighbors and radius 1 are generated. These histograms are then concatenated to build a 5900 dimensional feature vector. The dimension of this feature vector is then reduced by applying the principal component analysis (PCA) to it. Finally one-versus-all SVMs with RBF kernel are utilized to classify the data. For the emotion recognition task they perform the processing as described above on the whole face, followed by a majority voting over the estimates for each video. In case of the AU task, they instead detect upper face AUs on the upper half of the face image and lower face AUs on the lower half, respectively. Additionally, they filtered the training set such that for each video only one frame per AU combination was used for the training. For further details about the baseline system the interested reader is kindly referred to the challenge paper [17]. 4. Evaluation of the proposed system In this section, we describe in detail our proposed setup common to both of our systems for the two FERA2011 sub-challenges. This is followed by the setup and results of the emotion recognition and action unit detection subchallenge, respectively. Finally, the run-time is evaluated Common Setup In our proposed common framework, each frame of the video sequences is first processed by a face and eye detector. Afterwards, the face images are aligned with respect to the detected eye center locations so that the eye distance equals 31 pixels and the eyes are in the 26th row of the crop-out. The DCT is calculated for blocks of 8 8 pixels and only the first 10 coefficients are used. This leads to a feature vector of 800 dimensions. These parameters have shown to be the best in preliminary experiments on the GEMEP-FERA data using double cross-validation or a development and validation set splitting on the training data. Also, it was shown in [7] that a block size of 8 8 and keeping 10 coefficients gives the best compromise between recognition rate and data compression for face recognition. If no face is detected the frame is simply ignored. If no eyes are detected the frame is ignored for training, but for testing the unaligned face is processed as described above. For classification, we use the SVM implementation as provided by LIBSVM [6] using a RBF kernel. The best 3

4 Table 1. Confusion matrix for emotion recognition on the part of the test set with unknown persons (person independent). pred. \ truth Anger Fear Joy Relief Sadness Anger Fear Joy Relief Sadness values for the soft margin parameter C and kernel parameter γ are estimated by doing a grid search utilizing a 5-fold subject-independent cross-validation on the training data Emotion Recognition Sub-Challenge Based on this common setup we built a system for the emotion recognition sub-challenge, which will be described in detail in this subsection together with a discussion of the results on the GEMEP-FERA dataset Setup For the emotion recognition sub-challenge we use a single one-versus-all SVM classifier per emotion class. For each classifier a model for probability estimates is trained. For the training of such an emotion classifier we use all the frames of the videos labeled with the corresponding emotion as positive samples and all others as negative samples. The grid search is performed over C = 2 k and γ = 2 l with k = 3,..., 1 and l = 16,..., 7, respectively. In the classification stage each frame of the videos is first classified by all the emotion classifiers. The estimated emotion for that frame is then set to the emotion corresponding to the classifier with the highest probability. Since the task is to output one emotion estimate per video, we finally performed majority voting over all the emotion estimates of the frames for each video, like it is done in the baseline method [17]. In case the same number of estimates is returned by multiple classifiers, we select the emotion that is first in an alphabetical order Results The confusion matrix for the emotion task is shown in Table 1, Table 2 and Table 3 for the person independent and person specific portion of the test set as well as for the whole test set, respectively. The classification rates for our proposed DCT-based approach and the official LBP-based baseline are given in Table 4 1. We can see from the confusion matrix on the person specific portion in Table 2 that there is almost no confusion be- 1 The results slightly changed since the challenge submission, since we found a small bug in the meantime. The submitted system had 65.8%, 94.4%, and 77.3% person independent, person specific, and overall classification rate, respectively. Table 2. Confusion matrix for emotion recognition on the part of the test set with known persons (person specific). pred. \ truth Anger Fear Joy Relief Sadness Anger Fear Joy Relief Sadness Table 3. Confusion matrix for emotion recognition on the whole test set. pred. \ truth Anger Fear Joy Relief Sadness Anger Fear Joy Relief Sadness Table 4. Classification rates for emotion recognition using DCT and the LBP baseline [17]. Results are provided for the person specific (PS), person independent (PI), and overall partitions. DCT LBP Emotion PI PS Overall PI PS Overall anger 92.9% 100.0% 96.3% 86% 92% 89% fear 40.0% 90.0% 60.0% 7% 40% 20% joy 100.0% 100.0% 100.0% 70% 73% 71% relief 68.8% 100.0% 80.8% 31% 70% 46% sadness 40.0% 100.0% 64.0% 27% 90% 52% Average 68.3% 98.0% 80.2% 44% 73% 56% tween the emotions, which can also be seen from the person specific column of the classification results in Table 4, where 100% accuracy was achieved for all classes besides fear. This shows that the local appearance-based DCT face representation is very well suited for emotion classification on known subjects. On the person independent portion the results look different. Here, 100% accuracy is achieved only for joy, but compared to the person specific portion sadness is confused by a huge amount with anger. Also fear is almost equally confused with anger and joy. Looking at the overall performance, one can see that anger and joy seem to be relatively easy to distinguish from the other emotions, which can also be observed from the LBP-based baseline results, reproduced in Table 4. But emotions like fear and sadness, and to some extend also relief, are very challenging for both approaches, when the subject is unknown. This could also simply mean that subjects tend to express these emotions differently. But to get a statistically meaningful answer to that, one would have to evaluate on more data. It can be observed that for each of the portions the DCTbased approach gives around 24.2% absolute improvement over the LBP-based baseline approach. Furthermore a classification rate of 98% on the person specific test data portion shows that our approach is well suited especially for person specific emotion classification tasks. 4

5 4.3. Action Unit Detection Sub-Challenge In this subsection, we apply the same local appearancebased DCT face representation framework to the action unit detection sub-challenge and describe the setup, the differences to the emotion classification system as well as the results on the GEMEP-FERA dataset Setup For the action unit detection sub-challenge we also use a single one-versus-all SVM classifier per action unit. For the training of such an action unit classifier we select all the frames of the videos for which the corresponding action unit was labeled to be active as positive sample candidates and all others as negative sample candidates. Only for AU25 and AU26 we additionally remove all samples from the training set that are labeled with AD50 as being present. These sample candidates are then balanced by randomly removing samples from the class that is larger until we have the same amount of positive and negative samples. The grid search for the parameter estimation of the SVM is performed over C = k and γ = 2 l with k = 0,..., 31 and l = 15,..., 1, respectively. In the classification stage, each frame of the videos is independently classified by all the action unit classifiers and all action units for which the distance to the hyperplane is greater than 0 are set to be present in that frame. In case, frames have been ignored due to missing face detections, all action units are set to be inactive. Table 5. F1 scores for action unit detection using DCT and the LBP baseline [17]. Results are provided for the person specific (PS), person independent (PI), and overall partitions. Action DCT LBP Unit PI PS Overall PI PS Overall AU1 60.6% 30.7% 50.8% 63.3% 36.2% 56.7% AU2 52.0% 40.5% 47.9% 67.5% 40.0% 58.9% AU4 59.9% 52.6% 57.3% 13.3% 29.8% 19.2% AU6 82.5% 62.0% 76.1% 53.6% 25.5% 46.3% AU7 51.6% 57.1% 53.9% 49.3% 48.1% 48.9% AU % 53.9% 49.4% 44.5% 52.6% 47.9% AU % 82.9% 80.5% 76.9% 68.8% 74.2% AU % 22.8% 18.0% 8.2% 19.9% 13.3% AU % 24.8% 42.7% 37.8% 34.9% 36.9% AU % 26.8% 31.9% 12.6% 24.0% 17.6% AU % 76.5% 78.0% 79.6% 80.9% 80.2% AU % 40.8% 45.3% 37.1% 47.4% 41.5% Average 55.2% 47.6% 52.7% 45.3% 42.3% 45.1% Table 6. Two alternative forced choice (2AFC) score for action unit detection using DCT and the LBP baseline [17]. Results are provided for the person specific (PS), person independent (PI), and overall partitions. Action DCT LBP Unit PI PS Overall PI PS Overall AU1 52.6% 58.2% 58.2% 84.5% 61.3% 79.0% AU2 72.9% 64.7% 70.0% 81.8% 64.0% 76.7% AU4 51.8% 49.9% 51.1% 48.1% 60.7% 52.6% AU6 87.9% 81.5% 84.8% 69.0% 56.8% 65.7% AU7 68.5% 62.6% 66.5% 57.2% 53.0% 55.6% AU % 59.1% 53.4% 57.7% 62.7% 59.7% AU % 90.0% 84.2% 73.8% 70.0% 72.4% AU % 50.9% 49.8% 55.5% 56.7% 56.3% AU % 58.2% 60.7% 67.9% 66.1% 64.6% AU % 57.4% 70.7% 62.0% 59.9% 61.0% AU % 67.1% 66.9% 54.4% 66.9% 59.3% AU % 54.2% 59.4% 45.7% 55.5% 50.0% Average 65.3% 62.8% 64.6% 63.1% 61.1% 62.8% Results The F1 measures and two alternative forced choice (2AFC) scores for the action unit task using our proposed DCTbased approach and the official LBP-based baseline are shown in Table 5 and Table 6, respectively. 2 We can see in Table 5, that for AU6, AU12, and AU25 we achieve reasonable results, which shows that smiles are very well detectable. The poor performance for AU15, AU17, AU18, and AU26 could be due to the amount of data available for their training, since for these AUs we had only between 400 and 1000 samples, while for the other AUs there were around 1300 to 2700 samples available. But still our approach performs on most of them better than the baseline, which might indicate that using DCT allows you to have less training samples for a classifier to generalize. Since both systems show similar trends in performance, these results could be due to the distribution of the AUs in the dataset or simply a hint that these AUs are simply harder to detect in this dataset with these features. 2 The results slightly changed since the challenge submission, since we found a small bug in the meantime. The submitted system had 52.2% overall F1-score and 64% overall 2AFC. One interesting thing about both systems is that for about two third of the AUs the performance is better on the person independent test set portion. According to the organizers, this could be due to the distribution of AUs across subjects. Compared to the LBP-based baseline results, our approach achieves 7.6% absolute improvement in terms of the overall F1 score Runtime Our system has also proven to be very fast in classifying the test data. On an Intel Core i5-750 with 4 cores of 2.67 GHz each, it processed the 4733 frames of the action unit detection test set in approximately 3.35 minutes, which corresponds to a frame rate of around 23.6 frames per second. For the emotion recognition, the classification of the 7537 test frames took approximately 4.92 minutes, which corresponds to around 25.5 frames per second. Both of these timing tests include everything, from loading the videos from disk, over feature extraction to classifying the samples and finally saving the results back to disk. These results show that the system is real-time capable. 5

6 5. Conclusion and Future Work In this paper, we proposed a common framework for real-time action unit detection and emotion recognition using a local appearance-based DCT face representation and one-versus-all SVM classifiers. We evaluated the proposed approach within the FERA 2011 Challenge and compared its performance with the official LBP-based baseline results. We achieve 24.2% and 7.6% absolute improvement on the emotion recognition and action unit detection sub-challenge, respectively. This shows that the local appearance-based DCT face representation is well suited for emotion classification as well as action unit detection tasks. We also showed that the system runs in real-time. In the future, we plan to investigate the low performance of some of the action unit detectors as well as the generalization of the approach to other databases. In addition, incorporating time information and utilizing the locality information of the AUs, as well as some intelligent sample selection method for determining on which samples the detectors should be trained could improve the performance. 6. Acknowledgments This work is funded by the Concept for the Future of Karlsruhe Institute of Technology within the framework of the German Excellence Initiative. References [1] FERA2011 website. 2 [2] A. B. Ashraf, S. Lucey, J. F. Cohn, T. Chen, Z. Ambadar, K. M. Prkachin, and P. E. Solomon. The painful face Pain expression recognition using active appearance models. Image and Vision Computing, 27(12): , [3] T. Bänziger and K. R. Scherer. Introducing the Geneva Multimodal Emotion Portrayal (GEMEP) Corpus. In K. R. Scherer, T. Bänziger, and E. B. Roesch, editors, Blueprint for affective computing: A sourcebook, pages Oxford University Press, Oxford, England, [4] M. S. Bartlett, G. C. Littlewort, M. G. Frank, C. Lainscsek, I. R. Fasel, and J. R. Movellan. Automatic Recognition of Facial Actions in Spontaneous Expressions. Journal of Multimedia, , 2 [5] M. S. Bartlett and J. Whitehill. Automated facial expression measurement: Recent applications to basic research in human behavior, learning, and education. In A. Calder, G. Rhodes, J. V. Haxby, and M. H. Johnson, editors, Handbook of Face Perception. Oxford University Press, [6] C.-C. Chang and C.-J. Lin. LIBSVM: a library for support vector machines, Software available at http: // cjlin/libsvm. 3 [7] H. K. Ekenel. A Robust Face Recognition Algorithm for Real-World Applications. PhD thesis, Universität Karlsruhe (TH), Karlsruhe, Germany, Feb , 2, 3 [8] P. Ekman and W. V. Friesen. Facial Action Coding System: A Technique for the Measurement of Facial Movement. Consulting Psychologists Press, Palo Alto, California, [9] B. Fasel and J. Luettin. Automatic facial expression analysis: a survey. Pattern Recognition, 36(1): , Jan [10] C. Küblbeck and A. Ernst. Face detection and tracking in video sequences using the modified census transformation. Image and Vision Computing, 24(6): , June [11] P. Lucey, J. F. Cohn, T. Kanade, J. Saragih, and Z. Ambadar. The Extended Cohn-Kanade Dataset (CK+): A complete dataset for action unit and emotion-specified expression. In Proceedings of 3rd IEEE CVPR Workshop on CVPR for Human Communicative Behavior Analysis, , 2 [12] M. Madsen, R. el Kaliouby, M. Goodwin, and R. W. Picard. Technology for Just-In-Time In-Situ Learning of Facial Affect for Persons Diagnosed with an Autism Spectrum Disorder. In Proceedings of the 10th ACM Conference on Computers and Accessibility (ASSETS), [13] M. Pantic and L. Rothkrantz. Automatic analysis of facial expressions: The state of the art. IEEE Trans. on Pattern Analysis and Machine Intelligence, 22(12): , [14] Y. Saatci and C. Town. Cascaded Classification of Gender and Facial Expression using Active Appearance Models. In Proceedings of the 7th International Conference on Automatic Face and Gesture Recognition (FGR 06), [15] C. Shan, S. Gong, and P. W. McOwan. Facial expression recognition based on Local Binary Patterns: A comprehensive study. Image and Vision Computing, 27(6):803, , 2 [16] Y.-I. Tian, T. Kanade, and J. Cohn. Recognizing action units for facial expression analysis. IEEE Transactions on Pattern Analysis and Machine Intelligence, 23(2):97 115, [17] M. F. Valstar, B. Jiang, M. Méhu, M. Pantic, and K. Scherer. The First Facial Expression Recognition and Analysis Challenge. In Proc. of the IEEE International Conference on Automatic Face and Gesture Recognition, , 3, 4, 5 [18] E. Vural, M. Cetin, A. Ercil, G. Littlewort, M. Bartlett, and J. Movellan. Drowsy Driver Detection Through Facial Movement Analysis. In Proceedings of ICCV 2007 Workshop on Human Computer Interaction, [19] J. Whitehill, M. Bartlett, and J. Movellan. Automatic Facial Expression Recognition for Intelligent Tutoring Systems. In Proceedings of IEEE CVPR Workshop on CVPR for Human Communicative Behavior Analysis, [20] T. Wilhelm, H.-J. Böhme, and H.-M. Groß. Classification of Face Images for Gender, Age, Facial Expression, and Identity. In Proceedings of 15th International Conference on Artificial Neural Networks: Biological Inspirations (ICANN 2005), pages , [21] Z. Zeng, M. Pantic, G. I. Roisman, and T. S. Huang. A Survey of Affect Recognition Methods: Audio, Visual, and Spontaneous Expressions. IEEE Transactions on Pattern Analysis and Machine Intelligence, 31(1):39 58,

This is the accepted version of this article. To be published as : This is the author version published as:

This is the accepted version of this article. To be published as : This is the author version published as: QUT Digital Repository: http://eprints.qut.edu.au/ This is the author version published as: This is the accepted version of this article. To be published as : This is the author version published as: Chew,

More information

Facial expression recognition with spatiotemporal local descriptors

Facial expression recognition with spatiotemporal local descriptors Facial expression recognition with spatiotemporal local descriptors Guoying Zhao, Matti Pietikäinen Machine Vision Group, Infotech Oulu and Department of Electrical and Information Engineering, P. O. Box

More information

Emotion Recognition using a Cauchy Naive Bayes Classifier

Emotion Recognition using a Cauchy Naive Bayes Classifier Emotion Recognition using a Cauchy Naive Bayes Classifier Abstract Recognizing human facial expression and emotion by computer is an interesting and challenging problem. In this paper we propose a method

More information

Statistical and Neural Methods for Vision-based Analysis of Facial Expressions and Gender

Statistical and Neural Methods for Vision-based Analysis of Facial Expressions and Gender Proc. IEEE Int. Conf. on Systems, Man and Cybernetics (SMC 2004), Den Haag, pp. 2203-2208, IEEE omnipress 2004 Statistical and Neural Methods for Vision-based Analysis of Facial Expressions and Gender

More information

Study on Aging Effect on Facial Expression Recognition

Study on Aging Effect on Facial Expression Recognition Study on Aging Effect on Facial Expression Recognition Nora Algaraawi, Tim Morris Abstract Automatic facial expression recognition (AFER) is an active research area in computer vision. However, aging causes

More information

Automatic Facial Expression Recognition Using Boosted Discriminatory Classifiers

Automatic Facial Expression Recognition Using Boosted Discriminatory Classifiers Automatic Facial Expression Recognition Using Boosted Discriminatory Classifiers Stephen Moore and Richard Bowden Centre for Vision Speech and Signal Processing University of Surrey, Guildford, GU2 7JW,

More information

DEEP convolutional neural networks have gained much

DEEP convolutional neural networks have gained much Real-time emotion recognition for gaming using deep convolutional network features Sébastien Ouellet arxiv:8.37v [cs.cv] Aug 2 Abstract The goal of the present study is to explore the application of deep

More information

Facial Expression Analysis for Estimating Pain in Clinical Settings

Facial Expression Analysis for Estimating Pain in Clinical Settings Facial Expression Analysis for Estimating Pain in Clinical Settings Karan Sikka University of California San Diego 9450 Gilman Drive, La Jolla, California, USA ksikka@ucsd.edu ABSTRACT Pain assessment

More information

Facial Behavior as a Soft Biometric

Facial Behavior as a Soft Biometric Facial Behavior as a Soft Biometric Abhay L. Kashyap University of Maryland, Baltimore County 1000 Hilltop Circle, Baltimore, MD 21250 abhay1@umbc.edu Sergey Tulyakov, Venu Govindaraju University at Buffalo

More information

AUDIO-VISUAL EMOTION RECOGNITION USING AN EMOTION SPACE CONCEPT

AUDIO-VISUAL EMOTION RECOGNITION USING AN EMOTION SPACE CONCEPT 16th European Signal Processing Conference (EUSIPCO 28), Lausanne, Switzerland, August 25-29, 28, copyright by EURASIP AUDIO-VISUAL EMOTION RECOGNITION USING AN EMOTION SPACE CONCEPT Ittipan Kanluan, Michael

More information

Face Analysis : Identity vs. Expressions

Face Analysis : Identity vs. Expressions Hugo Mercier, 1,2 Patrice Dalle 1 Face Analysis : Identity vs. Expressions 1 IRIT - Université Paul Sabatier 118 Route de Narbonne, F-31062 Toulouse Cedex 9, France 2 Websourd Bâtiment A 99, route d'espagne

More information

The Extended Cohn-Kanade Dataset (CK+): A complete dataset for action unit and emotion-specified expression

The Extended Cohn-Kanade Dataset (CK+): A complete dataset for action unit and emotion-specified expression The Extended Cohn-Kanade Dataset (CK+): A complete dataset for action unit and emotion-specified expression Patrick Lucey 1,2, Jeffrey F. Cohn 1,2, Takeo Kanade 1, Jason Saragih 1, Zara Ambadar 2 Robotics

More information

Gender Based Emotion Recognition using Speech Signals: A Review

Gender Based Emotion Recognition using Speech Signals: A Review 50 Gender Based Emotion Recognition using Speech Signals: A Review Parvinder Kaur 1, Mandeep Kaur 2 1 Department of Electronics and Communication Engineering, Punjabi University, Patiala, India 2 Department

More information

Get The FACS Fast: Automated FACS face analysis benefits from the addition of velocity

Get The FACS Fast: Automated FACS face analysis benefits from the addition of velocity Get The FACS Fast: Automated FACS face analysis benefits from the addition of velocity Timothy R. Brick University of Virginia Charlottesville, VA 22904 tbrick@virginia.edu Michael D. Hunter University

More information

Facial Expression Recognition Using Principal Component Analysis

Facial Expression Recognition Using Principal Component Analysis Facial Expression Recognition Using Principal Component Analysis Ajit P. Gosavi, S. R. Khot Abstract Expression detection is useful as a non-invasive method of lie detection and behaviour prediction. However,

More information

A MULTIMODAL NONVERBAL HUMAN-ROBOT COMMUNICATION SYSTEM ICCB 2015

A MULTIMODAL NONVERBAL HUMAN-ROBOT COMMUNICATION SYSTEM ICCB 2015 VI International Conference on Computational Bioengineering ICCB 2015 M. Cerrolaza and S.Oller (Eds) A MULTIMODAL NONVERBAL HUMAN-ROBOT COMMUNICATION SYSTEM ICCB 2015 SALAH SALEH *, MANISH SAHU, ZUHAIR

More information

Utilizing Posterior Probability for Race-composite Age Estimation

Utilizing Posterior Probability for Race-composite Age Estimation Utilizing Posterior Probability for Race-composite Age Estimation Early Applications to MORPH-II Benjamin Yip NSF-REU in Statistical Data Mining and Machine Learning for Computer Vision and Pattern Recognition

More information

Facial Expression Biometrics Using Tracker Displacement Features

Facial Expression Biometrics Using Tracker Displacement Features Facial Expression Biometrics Using Tracker Displacement Features Sergey Tulyakov 1, Thomas Slowe 2,ZhiZhang 1, and Venu Govindaraju 1 1 Center for Unified Biometrics and Sensors University at Buffalo,

More information

Generalization of a Vision-Based Computational Model of Mind-Reading

Generalization of a Vision-Based Computational Model of Mind-Reading Generalization of a Vision-Based Computational Model of Mind-Reading Rana el Kaliouby and Peter Robinson Computer Laboratory, University of Cambridge, 5 JJ Thomson Avenue, Cambridge UK CB3 FD Abstract.

More information

A framework for the Recognition of Human Emotion using Soft Computing models

A framework for the Recognition of Human Emotion using Soft Computing models A framework for the Recognition of Human Emotion using Soft Computing models Md. Iqbal Quraishi Dept. of Information Technology Kalyani Govt Engg. College J Pal Choudhury Dept. of Information Technology

More information

HUMAN EMOTION DETECTION THROUGH FACIAL EXPRESSIONS

HUMAN EMOTION DETECTION THROUGH FACIAL EXPRESSIONS th June. Vol.88. No. - JATIT & LLS. All rights reserved. ISSN: -8 E-ISSN: 87- HUMAN EMOTION DETECTION THROUGH FACIAL EXPRESSIONS, KRISHNA MOHAN KUDIRI, ABAS MD SAID AND M YUNUS NAYAN Computer and Information

More information

Analysis of Emotion Recognition using Facial Expressions, Speech and Multimodal Information

Analysis of Emotion Recognition using Facial Expressions, Speech and Multimodal Information Analysis of Emotion Recognition using Facial Expressions, Speech and Multimodal Information C. Busso, Z. Deng, S. Yildirim, M. Bulut, C. M. Lee, A. Kazemzadeh, S. Lee, U. Neumann, S. Narayanan Emotion

More information

FERA Second Facial Expression Recognition and Analysis Challenge

FERA Second Facial Expression Recognition and Analysis Challenge FERA 2015 - Second Facial Expression Recognition and Analysis Challenge Michel F. Valstar 1, Timur Almaev 1, Jeffrey M. Girard 2, Gary McKeown 3, Marc Mehu 4, Lijun Yin 5, Maja Pantic 6,7 and Jeffrey F.

More information

Facial Expression Classification Using Convolutional Neural Network and Support Vector Machine

Facial Expression Classification Using Convolutional Neural Network and Support Vector Machine Facial Expression Classification Using Convolutional Neural Network and Support Vector Machine Valfredo Pilla Jr, André Zanellato, Cristian Bortolini, Humberto R. Gamba and Gustavo Benvenutti Borba Graduate

More information

Affective pictures and emotion analysis of facial expressions with local binary pattern operator: Preliminary results

Affective pictures and emotion analysis of facial expressions with local binary pattern operator: Preliminary results Affective pictures and emotion analysis of facial expressions with local binary pattern operator: Preliminary results Seppo J. Laukka 1, Antti Rantanen 1, Guoying Zhao 2, Matti Taini 2, Janne Heikkilä

More information

Emotion Affective Color Transfer Using Feature Based Facial Expression Recognition

Emotion Affective Color Transfer Using Feature Based Facial Expression Recognition , pp.131-135 http://dx.doi.org/10.14257/astl.2013.39.24 Emotion Affective Color Transfer Using Feature Based Facial Expression Recognition SeungTaek Ryoo and Jae-Khun Chang School of Computer Engineering

More information

Facial Action Unit Detection by Cascade of Tasks

Facial Action Unit Detection by Cascade of Tasks Facial Action Unit Detection by Cascade of Tasks Xiaoyu Ding Wen-Sheng Chu 2 Fernando De la Torre 2 Jeffery F. Cohn 2,3 Qiao Wang School of Information Science and Engineering, Southeast University, Nanjing,

More information

Detection of Facial Landmarks from Neutral, Happy, and Disgust Facial Images

Detection of Facial Landmarks from Neutral, Happy, and Disgust Facial Images Detection of Facial Landmarks from Neutral, Happy, and Disgust Facial Images Ioulia Guizatdinova and Veikko Surakka Research Group for Emotions, Sociality, and Computing Tampere Unit for Computer-Human

More information

IMPLEMENTATION OF AN AUTOMATED SMART HOME CONTROL FOR DETECTING HUMAN EMOTIONS VIA FACIAL DETECTION

IMPLEMENTATION OF AN AUTOMATED SMART HOME CONTROL FOR DETECTING HUMAN EMOTIONS VIA FACIAL DETECTION IMPLEMENTATION OF AN AUTOMATED SMART HOME CONTROL FOR DETECTING HUMAN EMOTIONS VIA FACIAL DETECTION Lim Teck Boon 1, Mohd Heikal Husin 2, Zarul Fitri Zaaba 3 and Mohd Azam Osman 4 1 Universiti Sains Malaysia,

More information

R Jagdeesh Kanan* et al. International Journal of Pharmacy & Technology

R Jagdeesh Kanan* et al. International Journal of Pharmacy & Technology ISSN: 0975-766X CODEN: IJPTFI Available Online through Research Article www.ijptonline.com FACIAL EMOTION RECOGNITION USING NEURAL NETWORK Kashyap Chiranjiv Devendra, Azad Singh Tomar, Pratigyna.N.Javali,

More information

A Vision-based Affective Computing System. Jieyu Zhao Ningbo University, China

A Vision-based Affective Computing System. Jieyu Zhao Ningbo University, China A Vision-based Affective Computing System Jieyu Zhao Ningbo University, China Outline Affective Computing A Dynamic 3D Morphable Model Facial Expression Recognition Probabilistic Graphical Models Some

More information

Real-time Automatic Deceit Detection from Involuntary Facial Expressions

Real-time Automatic Deceit Detection from Involuntary Facial Expressions Real-time Automatic Deceit Detection from Involuntary Facial Expressions Zhi Zhang, Vartika Singh, Thomas E. Slowe, Sergey Tulyakov, and Venugopal Govindaraju Center for Unified Biometrics and Sensors

More information

Facial Emotion Recognition with Facial Analysis

Facial Emotion Recognition with Facial Analysis Facial Emotion Recognition with Facial Analysis İsmail Öztel, Cemil Öz Sakarya University, Faculty of Computer and Information Sciences, Computer Engineering, Sakarya, Türkiye Abstract Computer vision

More information

Face Emotions and Short Surveys during Automotive Tasks

Face Emotions and Short Surveys during Automotive Tasks Face Emotions and Short Surveys during Automotive Tasks LEE QUINTANAR, PETE TRUJILLO, AND JEREMY WATSON March 2016 J.D. Power A Global Marketing Information Company jdpower.com Introduction Facial expressions

More information

EMOTION CLASSIFICATION: HOW DOES AN AUTOMATED SYSTEM COMPARE TO NAÏVE HUMAN CODERS?

EMOTION CLASSIFICATION: HOW DOES AN AUTOMATED SYSTEM COMPARE TO NAÏVE HUMAN CODERS? EMOTION CLASSIFICATION: HOW DOES AN AUTOMATED SYSTEM COMPARE TO NAÏVE HUMAN CODERS? Sefik Emre Eskimez, Kenneth Imade, Na Yang, Melissa Sturge- Apple, Zhiyao Duan, Wendi Heinzelman University of Rochester,

More information

Automatic Coding of Facial Expressions Displayed During Posed and Genuine Pain

Automatic Coding of Facial Expressions Displayed During Posed and Genuine Pain Automatic Coding of Facial Expressions Displayed During Posed and Genuine Pain Gwen C. Littlewort Machine Perception Lab, Institute for Neural Computation University of California, San Diego La Jolla,

More information

Emotion Detection Through Facial Feature Recognition

Emotion Detection Through Facial Feature Recognition Emotion Detection Through Facial Feature Recognition James Pao jpao@stanford.edu Abstract Humans share a universal and fundamental set of emotions which are exhibited through consistent facial expressions.

More information

A Deep Learning Approach for Subject Independent Emotion Recognition from Facial Expressions

A Deep Learning Approach for Subject Independent Emotion Recognition from Facial Expressions A Deep Learning Approach for Subject Independent Emotion Recognition from Facial Expressions VICTOR-EMIL NEAGOE *, ANDREI-PETRU BĂRAR *, NICU SEBE **, PAUL ROBITU * * Faculty of Electronics, Telecommunications

More information

Dimensional Emotion Prediction from Spontaneous Head Gestures for Interaction with Sensitive Artificial Listeners

Dimensional Emotion Prediction from Spontaneous Head Gestures for Interaction with Sensitive Artificial Listeners Dimensional Emotion Prediction from Spontaneous Head Gestures for Interaction with Sensitive Artificial Listeners Hatice Gunes and Maja Pantic Department of Computing, Imperial College London 180 Queen

More information

Local Image Structures and Optic Flow Estimation

Local Image Structures and Optic Flow Estimation Local Image Structures and Optic Flow Estimation Sinan KALKAN 1, Dirk Calow 2, Florentin Wörgötter 1, Markus Lappe 2 and Norbert Krüger 3 1 Computational Neuroscience, Uni. of Stirling, Scotland; {sinan,worgott}@cn.stir.ac.uk

More information

A Unified Probabilistic Framework For Measuring The Intensity of Spontaneous Facial Action Units

A Unified Probabilistic Framework For Measuring The Intensity of Spontaneous Facial Action Units A Unified Probabilistic Framework For Measuring The Intensity of Spontaneous Facial Action Units Yongqiang Li 1, S. Mohammad Mavadati 2, Mohammad H. Mahoor and Qiang Ji Abstract Automatic facial expression

More information

Face Emotions and Short Surveys during Automotive Tasks. April 2016

Face Emotions and Short Surveys during Automotive Tasks. April 2016 Face Emotions and Short Surveys during Automotive Tasks April 2016 Presented at the 2016 Council of American Survey Research Organizations (CASRO) Digital Conference, March 2016 A Global Marketing Information

More information

Deep Learning based FACS Action Unit Occurrence and Intensity Estimation

Deep Learning based FACS Action Unit Occurrence and Intensity Estimation Deep Learning based FACS Action Unit Occurrence and Intensity Estimation Amogh Gudi, H. Emrah Tasli, Tim M. den Uyl, Andreas Maroulis Vicarious Perception Technologies, Amsterdam, The Netherlands Abstract

More information

On Shape And the Computability of Emotions X. Lu, et al.

On Shape And the Computability of Emotions X. Lu, et al. On Shape And the Computability of Emotions X. Lu, et al. MICC Reading group 10.07.2013 1 On Shape and the Computability of Emotion X. Lu, P. Suryanarayan, R. B. Adams Jr., J. Li, M. G. Newman, J. Z. Wang

More information

Automatic detection of a driver s complex mental states

Automatic detection of a driver s complex mental states Automatic detection of a driver s complex mental states Zhiyi Ma 1, Marwa Mahmoud 2, Peter Robinson 2, Eduardo Dias 3, and Lee Skrypchuk 3 1 Department of Engineering, University of Cambridge, Cambridge,

More information

Recognizing Emotions from Facial Expressions Using Neural Network

Recognizing Emotions from Facial Expressions Using Neural Network Recognizing Emotions from Facial Expressions Using Neural Network Isidoros Perikos, Epaminondas Ziakopoulos, Ioannis Hatzilygeroudis To cite this version: Isidoros Perikos, Epaminondas Ziakopoulos, Ioannis

More information

Automatic Classification of Perceived Gender from Facial Images

Automatic Classification of Perceived Gender from Facial Images Automatic Classification of Perceived Gender from Facial Images Joseph Lemley, Sami Abdul-Wahid, Dipayan Banik Advisor: Dr. Razvan Andonie SOURCE 2016 Outline 1 Introduction 2 Faces - Background 3 Faces

More information

Discovering Facial Expressions for States of Amused, Persuaded, Informed, Sentimental and Inspired

Discovering Facial Expressions for States of Amused, Persuaded, Informed, Sentimental and Inspired Discovering Facial Expressions for States of Amused, Persuaded, Informed, Sentimental and Inspired Daniel McDuff Microsoft Research, Redmond, WA, USA This work was performed while at Affectiva damcduff@microsoftcom

More information

Advanced FACS Methodological Issues

Advanced FACS Methodological Issues 7/9/8 Advanced FACS Methodological Issues Erika Rosenberg, University of California, Davis Daniel Messinger, University of Miami Jeffrey Cohn, University of Pittsburgh The th European Conference on Facial

More information

EMOTION DETECTION THROUGH SPEECH AND FACIAL EXPRESSIONS

EMOTION DETECTION THROUGH SPEECH AND FACIAL EXPRESSIONS EMOTION DETECTION THROUGH SPEECH AND FACIAL EXPRESSIONS 1 KRISHNA MOHAN KUDIRI, 2 ABAS MD SAID AND 3 M YUNUS NAYAN 1 Computer and Information Sciences, Universiti Teknologi PETRONAS, Malaysia 2 Assoc.

More information

The Role of Face Parts in Gender Recognition

The Role of Face Parts in Gender Recognition The Role of Face Parts in Gender Recognition Yasmina Andreu Ramón A. Mollineda Pattern Analysis and Learning Section Computer Vision Group University Jaume I of Castellón (Spain) Y. Andreu, R.A. Mollineda

More information

COMPARISON BETWEEN GMM-SVM SEQUENCE KERNEL AND GMM: APPLICATION TO SPEECH EMOTION RECOGNITION

COMPARISON BETWEEN GMM-SVM SEQUENCE KERNEL AND GMM: APPLICATION TO SPEECH EMOTION RECOGNITION Journal of Engineering Science and Technology Vol. 11, No. 9 (2016) 1221-1233 School of Engineering, Taylor s University COMPARISON BETWEEN GMM-SVM SEQUENCE KERNEL AND GMM: APPLICATION TO SPEECH EMOTION

More information

NMF-Density: NMF-Based Breast Density Classifier

NMF-Density: NMF-Based Breast Density Classifier NMF-Density: NMF-Based Breast Density Classifier Lahouari Ghouti and Abdullah H. Owaidh King Fahd University of Petroleum and Minerals - Department of Information and Computer Science. KFUPM Box 1128.

More information

Using Affect Awareness to Modulate Task Experience: A Study Amongst Pre-Elementary School Kids

Using Affect Awareness to Modulate Task Experience: A Study Amongst Pre-Elementary School Kids Proceedings of the Twenty-Seventh International Florida Artificial Intelligence Research Society Conference Using Affect Awareness to Modulate Task Experience: A Study Amongst Pre-Elementary School Kids

More information

AUTOMATIC DETECTION AND INTENSITY ESTIMATION OF SPONTANEOUS SMILES. by Jeffrey M. Girard B.A. in Psychology/Philosophy, University of Washington, 2009

AUTOMATIC DETECTION AND INTENSITY ESTIMATION OF SPONTANEOUS SMILES. by Jeffrey M. Girard B.A. in Psychology/Philosophy, University of Washington, 2009 AUTOMATIC DETECTION AND INTENSITY ESTIMATION OF SPONTANEOUS SMILES by Jeffrey M. Girard B.A. in Psychology/Philosophy, University of Washington, 2009 Submitted to the Graduate Faculty of The Dietrich School

More information

Enhanced Facial Expressions Recognition using Modular Equable 2DPCA and Equable 2DPC

Enhanced Facial Expressions Recognition using Modular Equable 2DPCA and Equable 2DPC Enhanced Facial Expressions Recognition using Modular Equable 2DPCA and Equable 2DPC Sushma Choudhar 1, Sachin Puntambekar 2 1 Research Scholar-Digital Communication Medicaps Institute of Technology &

More information

Fusion of visible and thermal images for facial expression recognition

Fusion of visible and thermal images for facial expression recognition Front. Comput. Sci., 2014, 8(2): 232 242 DOI 10.1007/s11704-014-2345-1 Fusion of visible and thermal images for facial expression recognition Shangfei WANG 1,2, Shan HE 1,2,YueWU 3, Menghua HE 1,2,QiangJI

More information

Automated Tessellated Fundus Detection in Color Fundus Images

Automated Tessellated Fundus Detection in Color Fundus Images University of Iowa Iowa Research Online Proceedings of the Ophthalmic Medical Image Analysis International Workshop 2016 Proceedings Oct 21st, 2016 Automated Tessellated Fundus Detection in Color Fundus

More information

A Study of Facial Expression Reorganization and Local Binary Patterns

A Study of Facial Expression Reorganization and Local Binary Patterns A Study of Facial Expression Reorganization and Local Binary Patterns Poonam Verma #1, Deepshikha Rathore *2 #1 MTech Scholar,Sanghvi Innovative Academy Indore *2 Asst.Professor,Sanghvi Innovative Academy

More information

Facial Feature Model for Emotion Recognition Using Fuzzy Reasoning

Facial Feature Model for Emotion Recognition Using Fuzzy Reasoning Facial Feature Model for Emotion Recognition Using Fuzzy Reasoning Renan Contreras, Oleg Starostenko, Vicente Alarcon-Aquino, and Leticia Flores-Pulido CENTIA, Department of Computing, Electronics and

More information

Recognition of Facial Expressions for Images using Neural Network

Recognition of Facial Expressions for Images using Neural Network Recognition of Facial Expressions for Images using Neural Network Shubhangi Giripunje Research Scholar, Dept.of Electronics Engg., GHRCE, Nagpur, India Preeti Bajaj Senior IEEE Member, Professor, Dept.of

More information

Personalized Facial Attractiveness Prediction

Personalized Facial Attractiveness Prediction Personalized Facial Attractiveness Prediction Jacob Whitehill and Javier R. Movellan Machine Perception Laboratory University of California, San Diego La Jolla, CA 92093, USA {jake,movellan}@mplab.ucsd.edu

More information

Part III. Chapter 14 Insights on spontaneous facial expressions from automatic expression measurement

Part III. Chapter 14 Insights on spontaneous facial expressions from automatic expression measurement To appear in Giese,M. Curio, C., Bulthoff, H. (Eds.) Dynamic Faces: Insights from Experiments and Computation. MIT Press. 2009. Part III Chapter 14 Insights on spontaneous facial expressions from automatic

More information

Brain Tumor segmentation and classification using Fcm and support vector machine

Brain Tumor segmentation and classification using Fcm and support vector machine Brain Tumor segmentation and classification using Fcm and support vector machine Gaurav Gupta 1, Vinay singh 2 1 PG student,m.tech Electronics and Communication,Department of Electronics, Galgotia College

More information

Recognition of facial expressions using Gabor wavelets and learning vector quantization

Recognition of facial expressions using Gabor wavelets and learning vector quantization Engineering Applications of Artificial Intelligence 21 (2008) 1056 1064 www.elsevier.com/locate/engappai Recognition of facial expressions using Gabor wavelets and learning vector quantization Shishir

More information

Classroom Data Collection and Analysis using Computer Vision

Classroom Data Collection and Analysis using Computer Vision Classroom Data Collection and Analysis using Computer Vision Jiang Han Department of Electrical Engineering Stanford University Abstract This project aims to extract different information like faces, gender

More information

IEEE TRANSACTIONS ON AUDIO, SPEECH, AND LANGUAGE PROCESSING, VOL. 19, NO. 5, JULY

IEEE TRANSACTIONS ON AUDIO, SPEECH, AND LANGUAGE PROCESSING, VOL. 19, NO. 5, JULY IEEE TRANSACTIONS ON AUDIO, SPEECH, AND LANGUAGE PROCESSING, VOL. 19, NO. 5, JULY 2011 1057 A Framework for Automatic Human Emotion Classification Using Emotion Profiles Emily Mower, Student Member, IEEE,

More information

Identification of Neuroimaging Biomarkers

Identification of Neuroimaging Biomarkers Identification of Neuroimaging Biomarkers Dan Goodwin, Tom Bleymaier, Shipra Bhal Advisor: Dr. Amit Etkin M.D./PhD, Stanford Psychiatry Department Abstract We present a supervised learning approach to

More information

PERFORMANCE ANALYSIS OF THE TECHNIQUES EMPLOYED ON VARIOUS DATASETS IN IDENTIFYING THE HUMAN FACIAL EMOTION

PERFORMANCE ANALYSIS OF THE TECHNIQUES EMPLOYED ON VARIOUS DATASETS IN IDENTIFYING THE HUMAN FACIAL EMOTION PERFORMANCE ANALYSIS OF THE TECHNIQUES EMPLOYED ON VARIOUS DATASETS IN IDENTIFYING THE HUMAN FACIAL EMOTION Usha Mary Sharma 1, Jayanta Kumar Das 2, Trinayan Dutta 3 1 Assistant Professor, 2,3 Student,

More information

Emotion AI, Real-Time Emotion Detection using CNN

Emotion AI, Real-Time Emotion Detection using CNN Emotion AI, Real-Time Emotion Detection using CNN Tanner Gilligan M.S. Computer Science Stanford University tanner12@stanford.edu Baris Akis B.S. Computer Science Stanford University bakis@stanford.edu

More information

arxiv: v1 [cs.lg] 4 Feb 2019

arxiv: v1 [cs.lg] 4 Feb 2019 Machine Learning for Seizure Type Classification: Setting the benchmark Subhrajit Roy [000 0002 6072 5500], Umar Asif [0000 0001 5209 7084], Jianbin Tang [0000 0001 5440 0796], and Stefan Harrer [0000

More information

Automated facial expression measurement: Recent applications to basic research in human behavior, learning, and education

Automated facial expression measurement: Recent applications to basic research in human behavior, learning, and education 1 Automated facial expression measurement: Recent applications to basic research in human behavior, learning, and education Marian Stewart Bartlett and Jacob Whitehill, Institute for Neural Computation,

More information

Valence-arousal evaluation using physiological signals in an emotion recall paradigm. CHANEL, Guillaume, ANSARI ASL, Karim, PUN, Thierry.

Valence-arousal evaluation using physiological signals in an emotion recall paradigm. CHANEL, Guillaume, ANSARI ASL, Karim, PUN, Thierry. Proceedings Chapter Valence-arousal evaluation using physiological signals in an emotion recall paradigm CHANEL, Guillaume, ANSARI ASL, Karim, PUN, Thierry Abstract The work presented in this paper aims

More information

CPSC81 Final Paper: Facial Expression Recognition Using CNNs

CPSC81 Final Paper: Facial Expression Recognition Using CNNs CPSC81 Final Paper: Facial Expression Recognition Using CNNs Luis Ceballos Swarthmore College, 500 College Ave., Swarthmore, PA 19081 USA Sarah Wallace Swarthmore College, 500 College Ave., Swarthmore,

More information

Vital Responder: Real-time Health Monitoring of First- Responders

Vital Responder: Real-time Health Monitoring of First- Responders Vital Responder: Real-time Health Monitoring of First- Responders Ye Can 1,2 Advisors: Miguel Tavares Coimbra 2, Vijayakumar Bhagavatula 1 1 Department of Electrical & Computer Engineering, Carnegie Mellon

More information

TWO HANDED SIGN LANGUAGE RECOGNITION SYSTEM USING IMAGE PROCESSING

TWO HANDED SIGN LANGUAGE RECOGNITION SYSTEM USING IMAGE PROCESSING 134 TWO HANDED SIGN LANGUAGE RECOGNITION SYSTEM USING IMAGE PROCESSING H.F.S.M.Fonseka 1, J.T.Jonathan 2, P.Sabeshan 3 and M.B.Dissanayaka 4 1 Department of Electrical And Electronic Engineering, Faculty

More information

Age Estimation based on Multi-Region Convolutional Neural Network

Age Estimation based on Multi-Region Convolutional Neural Network Age Estimation based on Multi-Region Convolutional Neural Network Ting Liu, Jun Wan, Tingzhao Yu, Zhen Lei, and Stan Z. Li 1 Center for Biometrics and Security Research & National Laboratory of Pattern

More information

Bayesian Face Recognition Using Gabor Features

Bayesian Face Recognition Using Gabor Features Bayesian Face Recognition Using Gabor Features Xiaogang Wang, Xiaoou Tang Department of Information Engineering The Chinese University of Hong Kong Shatin, Hong Kong {xgwang1,xtang}@ie.cuhk.edu.hk Abstract

More information

Learning to Rank Authenticity from Facial Activity Descriptors Otto von Guericke University, Magdeburg - Germany

Learning to Rank Authenticity from Facial Activity Descriptors Otto von Guericke University, Magdeburg - Germany Learning to Rank Authenticity from Facial s Otto von Guericke University, Magdeburg - Germany Frerk Saxen, Philipp Werner, Ayoub Al-Hamadi The Task Real or Fake? Dataset statistics Training set 40 Subjects

More information

A Face-House Paradigm for Architectural Scene Analysis

A Face-House Paradigm for Architectural Scene Analysis A Face-House Paradigm for Architectural Scene Analysis Stephan K. Chalup Newcastle Robotics Lab School of Electrical Eng. and Computer Science The University of Newcastle NSW 2308 Australia stephan.chalup@newcastle.edu.au

More information

Assessment of Pain Using Facial Pictures Taken with a Smartphone

Assessment of Pain Using Facial Pictures Taken with a Smartphone 2015 IEEE 39th Annual International Computers, Software & Applications Conference Assessment of Pain Using Facial Pictures Taken with a Smartphone Mohammad Adibuzzaman 1, Colin Ostberg 1, Sheikh Ahamed

More information

ACTIVE APPEARANCE MODELS FOR AFFECT RECOGNITION USING FACIAL EXPRESSIONS. Matthew Stephen Ratliff

ACTIVE APPEARANCE MODELS FOR AFFECT RECOGNITION USING FACIAL EXPRESSIONS. Matthew Stephen Ratliff ACTIVE APPEARANCE MODELS FOR AFFECT RECOGNITION USING FACIAL EXPRESSIONS Matthew Stephen Ratliff A Thesis Submitted to the University of North Carolina Wilmington in Partial Fulfillment of the Requirements

More information

Audio-Visual Emotion Recognition in Adult Attachment Interview

Audio-Visual Emotion Recognition in Adult Attachment Interview Audio-Visual Emotion Recognition in Adult Attachment Interview Zhihong Zeng, Yuxiao Hu, Glenn I. Roisman, Zhen Wen, Yun Fu and Thomas S. Huang University of Illinois at Urbana-Champaign IBM T.J.Watson

More information

Affect Recognition for Interactive Companions

Affect Recognition for Interactive Companions Affect Recognition for Interactive Companions Ginevra Castellano School of Electronic Engineering and Computer Science Queen Mary University of London, UK ginevra@dcs.qmul.ac.uk Ruth Aylett School of Maths

More information

Apoptosis Detection for Adherent Cell Populations in Time-lapse Phase-contrast Microscopy Images

Apoptosis Detection for Adherent Cell Populations in Time-lapse Phase-contrast Microscopy Images Apoptosis Detection for Adherent Cell Populations in Time-lapse Phase-contrast Microscopy Images Seungil Huh 1, Dai Fei Elmer Ker 2, Hang Su 1, and Takeo Kanade 1 1 Robotics Institute, Carnegie Mellon

More information

Emotion Detection Using Physiological Signals. M.A.Sc. Thesis Proposal Haiyan Xu Supervisor: Prof. K.N. Plataniotis

Emotion Detection Using Physiological Signals. M.A.Sc. Thesis Proposal Haiyan Xu Supervisor: Prof. K.N. Plataniotis Emotion Detection Using Physiological Signals M.A.Sc. Thesis Proposal Haiyan Xu Supervisor: Prof. K.N. Plataniotis May 10 th, 2011 Outline Emotion Detection Overview EEG for Emotion Detection Previous

More information

A Study on Automatic Age Estimation using a Large Database

A Study on Automatic Age Estimation using a Large Database A Study on Automatic Age Estimation using a Large Database Guodong Guo WVU Guowang Mu NCCU Yun Fu BBN Technologies Charles Dyer UW-Madison Thomas Huang UIUC Abstract In this paper we study some problems

More information

Face Gender Classification on Consumer Images in a Multiethnic Environment

Face Gender Classification on Consumer Images in a Multiethnic Environment Face Gender Classification on Consumer Images in a Multiethnic Environment Wei Gao and Haizhou Ai Computer Science and Technology Department, Tsinghua University, Beijing 100084, China ahz@mail.tsinghua.edu.cn

More information

SmileMaze: A Tutoring System in Real-Time Facial Expression Perception and Production in Children with Autism Spectrum Disorder

SmileMaze: A Tutoring System in Real-Time Facial Expression Perception and Production in Children with Autism Spectrum Disorder SmileMaze: A Tutoring System in Real-Time Facial Expression Perception and Production in Children with Autism Spectrum Disorder Jeff Cockburn 1, Marni Bartlett 2, James Tanaka 1, Javier Movellan 2, Matt

More information

DISCRETE WAVELET PACKET TRANSFORM FOR ELECTROENCEPHALOGRAM- BASED EMOTION RECOGNITION IN THE VALENCE-AROUSAL SPACE

DISCRETE WAVELET PACKET TRANSFORM FOR ELECTROENCEPHALOGRAM- BASED EMOTION RECOGNITION IN THE VALENCE-AROUSAL SPACE DISCRETE WAVELET PACKET TRANSFORM FOR ELECTROENCEPHALOGRAM- BASED EMOTION RECOGNITION IN THE VALENCE-AROUSAL SPACE Farzana Kabir Ahmad*and Oyenuga Wasiu Olakunle Computational Intelligence Research Cluster,

More information

Using Computational Models to Understand ASD Facial Expression Recognition Patterns

Using Computational Models to Understand ASD Facial Expression Recognition Patterns Using Computational Models to Understand ASD Facial Expression Recognition Patterns Irene Feng Dartmouth College Computer Science Technical Report TR2017-819 May 30, 2017 Irene Feng 2 Literature Review

More information

ANALYSIS OF FACIAL FEATURES OF DRIVERS UNDER COGNITIVE AND VISUAL DISTRACTIONS

ANALYSIS OF FACIAL FEATURES OF DRIVERS UNDER COGNITIVE AND VISUAL DISTRACTIONS ANALYSIS OF FACIAL FEATURES OF DRIVERS UNDER COGNITIVE AND VISUAL DISTRACTIONS Nanxiang Li and Carlos Busso Multimodal Signal Processing (MSP) Laboratory Department of Electrical Engineering, The University

More information

From Dials to Facial Coding: Automated Detection of Spontaneous Facial Expressions for Media Research

From Dials to Facial Coding: Automated Detection of Spontaneous Facial Expressions for Media Research From Dials to Facial Coding: Automated Detection of Spontaneous Facial Expressions for Media Research Evan Kodra, Thibaud Senechal, Daniel McDuff, Rana el Kaliouby Abstract Typical consumer media research

More information

Hierarchical Age Estimation from Unconstrained Facial Images

Hierarchical Age Estimation from Unconstrained Facial Images Hierarchical Age Estimation from Unconstrained Facial Images STIC-AmSud Jhony Kaesemodel Pontes Department of Electrical Engineering Federal University of Paraná - Supervisor: Alessandro L. Koerich (/PUCPR

More information

Design of Palm Acupuncture Points Indicator

Design of Palm Acupuncture Points Indicator Design of Palm Acupuncture Points Indicator Wen-Yuan Chen, Shih-Yen Huang and Jian-Shie Lin Abstract The acupuncture points are given acupuncture or acupressure so to stimulate the meridians on each corresponding

More information

Sound Texture Classification Using Statistics from an Auditory Model

Sound Texture Classification Using Statistics from an Auditory Model Sound Texture Classification Using Statistics from an Auditory Model Gabriele Carotti-Sha Evan Penn Daniel Villamizar Electrical Engineering Email: gcarotti@stanford.edu Mangement Science & Engineering

More information

Estimating smile intensity: A better way

Estimating smile intensity: A better way Pattern Recognition Letters journal homepage: www.elsevier.com Estimating smile intensity: A better way Jeffrey M. Girard a,, Jeffrey F. Cohn a,b, Fernando De la Torre b a Department of Psychology, University

More information

Gene Selection for Tumor Classification Using Microarray Gene Expression Data

Gene Selection for Tumor Classification Using Microarray Gene Expression Data Gene Selection for Tumor Classification Using Microarray Gene Expression Data K. Yendrapalli, R. Basnet, S. Mukkamala, A. H. Sung Department of Computer Science New Mexico Institute of Mining and Technology

More information

Development of novel algorithm by combining Wavelet based Enhanced Canny edge Detection and Adaptive Filtering Method for Human Emotion Recognition

Development of novel algorithm by combining Wavelet based Enhanced Canny edge Detection and Adaptive Filtering Method for Human Emotion Recognition International Journal of Engineering Research and Development e-issn: 2278-067X, p-issn: 2278-800X, www.ijerd.com Volume 12, Issue 9 (September 2016), PP.67-72 Development of novel algorithm by combining

More information

Towards Multimodal Emotion Recognition: A New Approach

Towards Multimodal Emotion Recognition: A New Approach Towards Multimodal Emotion Recognition: A New Approach Marco Paleari TEleRobotics and Applications Italian Institute of Technology Genoa, Italy marco.paleari @ iit.it Benoit Huet Multimedia Department

More information