Recognition of Facial Expressions with Autoencoders and Convolutional-Nets

Les humains communiquent via différents types de canaux: les mots, la voix, les gestes du corps, des émotions, etc. Pour cette raison, un ordinateur doit percevoir ces divers canaux de communication pour pouvoir interagir intelligemment avec les humains, par exemple en faisant usage de microphones et de webcams. Dans cette thèse, nous nous intéressons à déterminer les émotions humaines à partir d’images ou de vidéo de visages afin d’ensuite utiliser ces informations dans différents domaines d’applications. Ce mémoire débute par une brève introduction à l'apprentissage machine en s’attardant aux modèles et algorithmes que nous avons utilisés tels que les perceptrons multicouches, réseaux de neurones à convolution et autoencodeurs. Elle présente ensuite les résultats de l'application de ces modèles sur plusieurs ensembles de données d'expressions et émotions faciales. Nous nous concentrons sur l'étude des différents types d’autoencodeurs (autoencodeur débruitant, autoencodeur contractant, etc) afin de révéler certaines de leurs limitations, comme la possibilité d'obtenir de la coadaptation entre les filtres ou encore d’obtenir une courbe spectrale trop lisse, et étudions de nouvelles idées pour répondre à ces problèmes. Nous proposons également une nouvelle approche pour surmonter une limite des autoencodeurs traditionnellement entrainés de façon purement non-supervisée, c'est-à-dire sans utiliser aucune connaissance de la tâche que nous voulons finalement résoudre (comme la prévision des étiquettes de classe) en développant un nouveau critère d'apprentissage semi-supervisé qui exploite un faible nombre de données étiquetées en combinaison avec une grande quantité de données non-étiquetées afin d'apprendre une représentation adaptée à la tâche de classification, et d'obtenir une meilleure performance de classification. Finalement, nous décrivons le fonctionnement général de notre système de détection d'émotions et proposons de nouvelles idées pouvant mener à de futurs travaux.

El debate sobre la memoria del conflicto armado en el Perú (1980-2000) : un acercamiento a través de la cultura visual.

À travers d'un regard de la culture visuelle, ce mémoire de maîtrise explorera le débat entourant la mémoire du conflit armé au Pérou (1980-2000), suite à la présentation de l’Informe Final (2003) par la Commission de la vérité et de la réconciliation. Tout d’abord, nous tracerons un portrait des débats qui ont eu lieu dans la sphère politique pour ensuite nous concentrer sur la polémique entourant le Musée de la mémoire et le monument Ojo que llora. Par la suite, nous explorerons la représentation visuelle du conflit à travers l’exposition de photos Yuyanapaq (2004), les films La boca del lobo (1988), La vida es una sola (1993) et Sangre Inocente (2000), en établissant un dialogue avec les études sur la mémoire et la critique académique. Après avoir dressé le tableau des imaginaires visuels du conflit, nous nous attarderons sur l’étape postconflit (2000 à aujourd’hui) pour aborder le débat entourant la postmémoire dans les films Madeinusa (2005) et La teta asustada (2009). Nous croyons que les positions entourant ce débat et la représentation du conflit dans la culture visuelle suggèrent que les mémoires sont toujours en opposition encore aujourd'hui au Pérou. Certaines réaffirment un discours hégémonique, comme l’Informe de Uchuraccay (1983), alors que d’autres s’opposent aux visions totalitaires de la mémoire; pensons notamment à l’émergence d’un nouveau cinéma de région qui implique les populations touchées par la violence dans la production et la projection de ses propres visions du conflit.

Concept oriented biomedical information retrieval

Le domaine biomédical est probablement le domaine où il y a les ressources les plus riches. Dans ces ressources, on regroupe les différentes expressions exprimant un concept, et définit des relations entre les concepts. Ces ressources sont construites pour faciliter l’accès aux informations dans le domaine. On pense généralement que ces ressources sont utiles pour la recherche d’information biomédicale. Or, les résultats obtenus jusqu’à présent sont mitigés : dans certaines études, l’utilisation des concepts a pu augmenter la performance de recherche, mais dans d’autres études, on a plutôt observé des baisses de performance. Cependant, ces résultats restent difficilement comparables étant donné qu’ils ont été obtenus sur des collections différentes. Il reste encore une question ouverte si et comment ces ressources peuvent aider à améliorer la recherche d’information biomédicale. Dans ce mémoire, nous comparons les différentes approches basées sur des concepts dans un même cadre, notamment l’approche utilisant les identificateurs de concept comme unité de représentation, et l’approche utilisant des expressions synonymes pour étendre la requête initiale. En comparaison avec l’approche traditionnelle de "sac de mots", nos résultats d’expérimentation montrent que la première approche dégrade toujours la performance, mais la seconde approche peut améliorer la performance. En particulier, en appariant les expressions de concepts comme des syntagmes stricts ou flexibles, certaines méthodes peuvent apporter des améliorations significatives non seulement par rapport à la méthode de "sac de mots" de base, mais aussi par rapport à la méthode de Champ Aléatoire Markov (Markov Random Field) qui est une méthode de l’état de l’art dans le domaine. Ces résultats montrent que quand les concepts sont utilisés de façon appropriée, ils peuvent grandement contribuer à améliorer la performance de recherche d’information biomédicale. Nous avons participé au laboratoire d’évaluation ShARe/CLEF 2014 eHealth. Notre résultat était le meilleur parmi tous les systèmes participants.

Design And Development Of A Named Entity Based Question Answering System For Malayalam Language

This is a Named Entity Based Question Answering System for Malayalam Language. Although a vast amount of information is available today in digital form, no effective information access mechanism exists to provide humans with convenient information access. Information Retrieval and Question Answering systems are the two mechanisms available now for information access. Information systems typically return a long list of documents in response to a user’s query which are to be skimmed by the user to determine whether they contain an answer. But a Question Answering System allows the user to state his/her information need as a natural language question and receives most appropriate answer in a word or a sentence or a paragraph. This system is based on Named Entity Tagging and Question Classification. Document tagging extracts useful information from the documents which will be used in finding the answer to the question. Question Classification extracts useful information from the question to determine the type of the question and the way in which the question is to be answered. Various Machine Learning methods are used to tag the documents. Rule-Based Approach is used for Question Classification. Malayalam belongs to the Dravidian family of languages and is one of the four major languages of this family. It is one of the 22 Scheduled Languages of India with official language status in the state of Kerala. It is spoken by 40 million people. Malayalam is a morphologically rich agglutinative language and relatively of free word order. Also Malayalam has a productive morphology that allows the creation of complex words which are often highly ambiguous. Document tagging tools such as Parts-of-Speech Tagger, Phrase Chunker, Named Entity Tagger, and Compound Word Splitter are developed as a part of this research work. No such tools were available for Malayalam language. Finite State Transducer, High Order Conditional Random Field, Artificial Immunity System Principles, and Support Vector Machines are the techniques used for the design of these document preprocessing tools. This research work describes how the Named Entity is used to represent the documents. Single sentence questions are used to test the system. Overall Precision and Recall obtained are 88.5% and 85.9% respectively. This work can be extended in several directions. The coverage of non-factoid questions can be increased and also it can be extended to include open domain applications. Reference Resolution and Word Sense Disambiguation techniques are suggested as the future enhancements

Handling OOV Words in Phrase-Based Statistical Machine Translation for Malayalam

Statistical Machine Translation (SMT) is one of the potential applications in the field of Natural Language Processing. The translation process in SMT is carried out by acquiring translation rules automatically from the parallel corpora. However, for many language pairs (e.g. Malayalam- English), they are available only in very limited quantities. Therefore, for these language pairs a huge portion of phrases encountered at run-time will be unknown. This paper focuses on methods for handling such out-of-vocabulary (OOV) words in Malayalam that cannot be translated to English using conventional phrase-based statistical machine translation systems. The OOV words in the source sentence are pre-processed to obtain the root word and its suffix. Different inflected forms of the OOV root are generated and a match is looked up for the word variants in the phrase translation table of the translation model. A Vocabulary filter is used to choose the best among the translations of these word variants by finding the unigram count. A match for the OOV suffix is also looked up in the phrase entries and the target translations are filtered out. Structuring of the filtered phrases is done and SMT translation model is extended by adding OOV with its new phrase translations. By the results of the manual evaluation done it is observed that amount of OOV words in the input has been reduced considerably

Extension schemes for the Alignment Model of English-Malayalam Statistical Machine Translator

In Statistical Machine Translation from English to Malayalam, an unseen English sentence is translated into its equivalent Malayalam sentence using statistical models. A parallel corpus of English-Malayalam is used in the training phase. Word to word alignments has to be set among the sentence pairs of the source and target language before subjecting them for training. This paper deals with certain techniques which can be adopted for improving the alignment model of SMT. Methods to incorporate the parts of speech information into the bilingual corpus has resulted in eliminating many of the insignificant alignments. Also identifying the name entities and cognates present in the sentence pairs has proved to be advantageous while setting up the alignments. Presence of Malayalam words with predictable translations has also contributed in reducing the insignificant alignments. Moreover, reduction of the unwanted alignments has brought in better training results. Experiments conducted on a sample corpus have generated reasonably good Malayalam translations and the results are verified with F measure, BLEU and WER evaluation metrics.

Statistical evaluation of texture analysis from the biocrystallization method

The consumers are becoming more concerned about food quality, especially regarding how, when and where the foods are produced (Haglund et al., 1999; Kahl et al., 2004; Alföldi, et al., 2006). Therefore, during recent years there has been a growing interest in the methods for food quality assessment, especially in the picture-development methods as a complement to traditional chemical analysis of single compounds (Kahl et al., 2006). The biocrystallization as one of the picture-developing method is based on the crystallographic phenomenon that when crystallizing aqueous solutions of dihydrate CuCl2 with adding of organic solutions, originating, e.g., from crop samples, biocrystallograms are generated with reproducible crystal patterns (Kleber & Steinike-Hartung, 1959). Its output is a crystal pattern on glass plates from which different variables (numbers) can be calculated by using image analysis. However, there is a lack of a standardized evaluation method to quantify the morphological features of the biocrystallogram image. Therefore, the main sakes of this research are (1) to optimize an existing statistical model in order to describe all the effects that contribute to the experiment, (2) to investigate the effect of image parameters on the texture analysis of the biocrystallogram images, i.e., region of interest (ROI), color transformation and histogram matching on samples from the project 020E170/F financed by the Federal Ministry of Food, Agriculture and Consumer Protection(BMELV).The samples are wheat and carrots from controlled field and farm trials, (3) to consider the strongest effect of texture parameter with the visual evaluation criteria that have been developed by a group of researcher (University of Kassel, Germany; Louis Bolk Institute (LBI), Netherlands and Biodynamic Research Association Denmark (BRAD), Denmark) in order to clarify how the relation of the texture parameter and visual characteristics on an image is. The refined statistical model was accomplished by using a lme model with repeated measurements via crossed effects, programmed in R (version 2.1.0). The validity of the F and P values is checked against the SAS program. While getting from the ANOVA the same F values, the P values are bigger in R because of the more conservative approach. The refined model is calculating more significant P values. The optimization of the image analysis is dealing with the following parameters: ROI(Region of Interest which is the area around the geometrical center), color transformation (calculation of the 1 dimensional gray level value out of the three dimensional color information of the scanned picture, which is necessary for the texture analysis), histogram matching (normalization of the histogram of the picture to enhance the contrast and to minimize the errors from lighting conditions). The samples were wheat from DOC trial with 4 field replicates for the years 2003 and 2005, “market samples”(organic and conventional neighbors with the same variety) for 2004 and 2005, carrot where the samples were obtained from the University of Kassel (2 varieties, 2 nitrogen treatments) for the years 2004, 2005, 2006 and “market samples” of carrot for the years 2004 and 2005. The criterion for the optimization was repeatability of the differentiation of the samples over the different harvest(years). For different samples different ROIs were found, which reflect the different pictures. The best color transformation that shows efficiently differentiation is relied on gray scale, i.e., equal color transformation. The second dimension of the color transformation only appeared in some years for the effect of color wavelength(hue) for carrot treated with different nitrate fertilizer levels. The best histogram matching is the Gaussian distribution. The approach was to find a connection between the variables from textural image analysis with the different visual criteria. The relation between the texture parameters and visual evaluation criteria was limited to the carrot samples, especially, as it could be well differentiated by the texture analysis. It was possible to connect groups of variables of the texture analysis with groups of criteria from the visual evaluation. These selected variables were able to differentiate the samples but not able to classify the samples according to the treatment. Contrarily, in case of visual criteria which describe the picture as a whole there is a classification in 80% of the sample cases possible. Herewith, it clearly can find the limits of the single variable approach of the image analysis (texture analysis).

Geometry and Photometry in 3D Visual Recognition

The report addresses the problem of visual recognition under two sources of variability: geometric and photometric. The geometric deals with the relation between 3D objects and their views under orthographic and perspective projection. The photometric deals with the relation between 3D matte objects and their images under changing illumination conditions. Taken together, an alignment-based method is presented for recognizing objects viewed from arbitrary viewing positions and illuminated by arbitrary settings of light sources.

Geometric and Algebraic Aspects of 3D Affine and Projective Structures from Perspective 2D Views

We investigate the differences --- conceptually and algorithmically --- between affine and projective frameworks for the tasks of visual recognition and reconstruction from perspective views. It is shown that an affine invariant exists between any view and a fixed view chosen as a reference view. This implies that for tasks for which a reference view can be chosen, such as in alignment schemes for visual recognition, projective invariants are not really necessary. We then use the affine invariant to derive new algebraic connections between perspective views. It is shown that three perspective views of an object are connected by certain algebraic functions of image coordinates alone (no structure or camera geometry needs to be involved).

Translation Invariance in Object Recognition, and Its Relation to Other Visual Transformations

Human object recognition is generally considered to tolerate changes of the stimulus position in the visual field. A number of recent studies, however, have cast doubt on the completeness of translation invariance. In a new series of experiments we tried to investigate whether positional specificity of short-term memory is a general property of visual perception. We tested same/different discrimination of computer graphics models that were displayed at the same or at different locations of the visual field, and found complete translation invariance, regardless of the similarity of the animals and irrespective of direction and size of the displacement (Exp. 1 and 2). Decisions were strongly biased towards same decisions if stimuli appeared at a constant location, while after translation subjects displayed a tendency towards different decisions. Even if the spatial order of animal limbs was randomized ("scrambled animals"), no deteriorating effect of shifts in the field of view could be detected (Exp. 3). However, if the influence of single features was reduced (Exp. 4 and 5) small but significant effects of translation could be obtained. Under conditions that do not reveal an influence of translation, rotation in depth strongly interferes with recognition (Exp. 6). Changes of stimulus size did not reduce performance (Exp. 7). Tolerance to these object transformations seems to rely on different brain mechanisms, with translation and scale invariance being achieved in principle, while rotation invariance is not.

Face processing in humans is compatible with a simple shape-based model of vision

Understanding how the human visual system recognizes objects is one of the key challenges in neuroscience. Inspired by a large body of physiological evidence (Felleman and Van Essen, 1991; Hubel and Wiesel, 1962; Livingstone and Hubel, 1988; Tso et al., 2001; Zeki, 1993), a general class of recognition models has emerged which is based on a hierarchical organization of visual processing, with succeeding stages being sensitive to image features of increasing complexity (Hummel and Biederman, 1992; Riesenhuber and Poggio, 1999; Selfridge, 1959). However, these models appear to be incompatible with some well-known psychophysical results. Prominent among these are experiments investigating recognition impairments caused by vertical inversion of images, especially those of faces. It has been reported that faces that differ "featurally" are much easier to distinguish when inverted than those that differ "configurally" (Freire et al., 2000; Le Grand et al., 2001; Mondloch et al., 2002) ??finding that is difficult to reconcile with the aforementioned models. Here we show that after controlling for subjects' expectations, there is no difference between "featurally" and "configurally" transformed faces in terms of inversion effect. This result reinforces the plausibility of simple hierarchical models of object representation and recognition in cortex.

Robust decoupled visual servoing based on structured light

This paper focuses on the problem of realizing a plane-to-plane virtual link between a camera attached to the end-effector of a robot and a planar object. In order to do the system independent to the object surface appearance, a structured light emitter is linked to the camera so that 4 laser pointers are projected onto the object. In a previous paper we showed that such a system has good performance and nice characteristics like partial decoupling near the desired state and robustness against misalignment of the emitter and the camera (J. Pages et al., 2004). However, no analytical results concerning the global asymptotic stability of the system were obtained due to the high complexity of the visual features utilized. In this work we present a better set of visual features which improves the properties of the features in (J. Pages et al., 2004) and for which it is possible to prove the global asymptotic stability

Plane-to-plane positioning from image-based visual servoing and structured light

In this paper we face the problem of positioning a camera attached to the end-effector of a robotic manipulator so that it gets parallel to a planar object. Such problem has been treated for a long time in visual servoing. Our approach is based on linking to the camera several laser pointers so that its configuration is aimed to produce a suitable set of visual features. The aim of using structured light is not only for easing the image processing and to allow low-textured objects to be treated, but also for producing a control scheme with nice properties like decoupling, stability, well conditioning and good camera trajectory

Factores pronósticos en la agudeza visual final de pacientes con oclusión venosa retiniana tratados con terapia antiangiogénica

Objetivo: determinar los diferentes factores clínicos y de imagen pronósticos para la agudeza visual final a los 3, 6 y 12 meses en los pacientes con oclusiones venosas retinianas tratados con terapia antiangiogénica. Material y métodos: estudio longitudinal de una cohorte de 60 pacientes con oclusión venosa retiniana tratados con terapia antiangiogénica intravitrea con ranibizumab y bevacizumab tomados retrospectivamente entre 2010 y 2012. Posteriormente se realizó un estudio analítico de cohorte retrospectiva, para establecer la asociación y predicción de los cambios de la agudeza visual. Resultados: se encontró una diferencia estadísticamente significativa entre la línea base, 3, 6 y 12 meses post tratamiento. (p < 0.001, prueba de Friedman). En el modelo explicativo de los cambios de la agudeza visual a los 3, 6 y 12 meses con la línea base, se encontró asociación significativa la presencia de IS/OS a los 3 meses con los cambios de la agudeza visual a los 6 meses (r2=.232, p< 0.001) y a los 12 meses (r2=.506, p< 0.001), en este último también se encontró asociado el género masculino (r2=.277 p< 0.001). Tomando los valores de la agudeza visual a los 3,6 y 12 meses y como covariable la agudeza visual inicial se encontró asociación significativa en los tres momentos, 3 meses (r2=0.697, p< 0.001), 6 meses (r2=745, p< 0.001) y 12 meses y (r2=786, p< 0.001), se encontró asociación significativa el presentar oclusión venosa de rama comparada con oclusión central con los cambios de la agudeza visual a los 6 meses (r2=.662, p=0.04) Conclusión: la presencia de IS/OS a los 3 meses y la agudeza visual inicial tienen una alta capacidad predictora de la agudeza visual a los 6 y 12 meses, el género masculino tiene relación con mejoría de la agudeza visual final y la oclusión de rama tiene relación con la mejoría de la agudeza visual a 6 meses.

Hiperautofluorescencia foveal como factor predictor de recuperación visual en pacientes con desprendimiento regmatógeno de retina

Introducción: una de las causas de pobre ganancia visual luego de un tratamiento exitoso de desprendimiento de retina, sin complicaciones, es el daño de los fotoreceptores, reflejada en una disrupción de la capa de la zona elipsoide y membrana limitante externa (MLE). En otras patologías se ha demostrado que la hiperautofluorescencia foveal se correlaciona con la integridad de la zona elipsoide y MLE y una mejor recuperación visual. Objetivos: evaluar la asociación entre la hiperautofluorescencia foveal, la integridad de la capa de la zona elipsoide y recuperación visual luego de desprendimiento de retina regmatógeno (DRR) exitosamente tratado. Evaluar la concordancia inter-evaluador de estos exámenes. Metodología: estudio de corte transversal de autofluorescencia foveal y tomografía óptica coherente macular de dominio espectral en 65 pacientes con DRR evaluados por 3 evaluadores independientes. La concordancia inter-evaluador se estudio mediante Kappa de Cohen y la asociación entre las diferentes variables mediante la prueba chi cuadrado y pruebas Z para comparación de proporciones. Resultados: La concordancia de la autofluorescencia fue razonable y la de la tomografía óptica coherente macular buena a muy buena. Sujetos que presentaron hiperautofluorescencia foveal asociada a integridad de la capa de la zona elipsoide tuvieron 20% más de posibilidad de recuperar agudeza visual final mejor a 20/50 que los que no cumplieron éstas características. Conclusión: Existe una asociación clínicamente importante entre la hiperautofluorescencia foveal, la integridad de la capa de zona elipsoide y la mejor agudeza visual final, sin embargo ésta no fue estadísticamente significativa (p=0.39)