This thesis researches techniques for modelling static facial expressions, as well as the dynamics of continuous facial motion. We demonstrate how static and dynamic properties of facial expressions can be represented within a linear and nonlinear context, respectively. These two representations do not act in isolation, but are mutually reinforcing in conceding a cohesive framework for the analysis, animation, and manipulation of expressive faces. We derive a basis for the linear space of expressions through Principal Components Analysis (PCA). We introduce and formalize the notion of "expression manifolds", manifolds residing in PCA space that model motion dynamics for semantically similar expressions. We then integrate these manifolds into an animation workflow by performing Nonlinear Dimensionality Reduction (NLDR) on the expression manifolds. This operation yields expression maps that encode a wealth of information relating to complex facial dynamics, in a low dimensional space that is intuitive to navigate and efficient to manage.

This thesis researches techniques for modelling static facial expressions, as well as the dynamics of continuous facial motion. We demonstrate how static and dynamic properties of facial expressions can be represented within a linear and nonlinear context, respectively. These two representations do not act in isolation, but are mutually reinforcing in conceding a cohesive framework for the analysis, animation, and manipulation of expressive faces. We derive a basis for the linear space of expressions through Principal Components Analysis (PCA). We introduce and formalize the notion of "expression manifolds", manifolds residing in PCA space that model motion dynamics for semantically similar expressions. We then integrate these manifolds into an animation workflow by performing Nonlinear Dimensionality Reduction (NLDR) on the expression manifolds. This operation yields expression maps that encode a wealth of information relating to complex facial dynamics, in a low dimensional space that is intuitive to navigate and efficient to manage.

Le mouvement de la marche est un processus essentiel de l'activité humaine et aussi le résultat de nombreuses interactions collaboratives entre les systèmes neurologiques, articulaires et musculo-squelettiques fonctionnant ensemble efficacement. Ceci explique pourquoi une analyse de la marche est aujourd'hui de plus en plus utilisée pour le diagnostic (et aussi la prévention) de différents types de maladies (neurologiques, musculaires, orthopédique, etc.). Ce rapport présente une nouvelle méthode pour visualiser rapidement les différentes parties du corps humain liées à une possible asymétrie (temporellement invariante par translation) existant dans la démarche d'un patient pour une possible utilisation clinique quotidienne. L'objectif est de fournir une méthode à la fois facile et peu dispendieuse permettant la mesure et l'affichage visuel, d'une manière intuitive et perceptive, des différentes parties asymétriques d'une démarche. La méthode proposée repose sur l'utilisation d'un capteur de profondeur peu dispendieux (la Kinect) qui est très bien adaptée pour un diagnostique rapide effectué dans de petites salles médicales car ce capteur est d'une part facile à installer et ne nécessitant aucun marqueur. L'algorithme que nous allons présenter est basé sur le fait que la marche saine possède des propriétés de symétrie (relativement à une invariance temporelle) dans le plan coronal. / The gait movement is an essential process of the human activity and also the result of coordinated effort between the neurological, articular and musculoskeletal systems. This motivates why gait analysis is important and also increasingly used nowadays for the (possible early) diagnosis of many different types (neurological, muscular, orthopedic, etc.) of diseases. This paper introduces a novel method to quickly visualize the different parts of the body related to an asymmetric movement in the human gait of a patient for daily clinical. The goal is to provide a cheap and easy-to-use method to measure the gait asymmetry and display results in a perceptually relevant manner. This method relies on an affordable consumer depth sensor, the Kinect. The Kinect was chosen because this device is amenable for use in small, confined area, like a living room. Also, since it is marker-less, it provides a fast non-invasive diagnostic. The algorithm we are going to introduce relies on the fact that a healthy walk has (temporally shift-invariant) symmetry properties in the coronal plane.

Variational Auto-Encoders (VAEs) are one of the most famous deep generative models. After showing that standard VAEs may not preserve the topology, that is the shape of the data, between the input and the latent space, we tried to modify them so that the topology is preserved. This would help in particular for performing interpolations in the latent space. Our main contribution is two folds. Firstly, we propose successfully the InvMap-VAE which is a simple way to turn any dimensionality reduction technique, given its embedding, into a generative model within a VAE framework providing an inverse mapping, with all the advantages that this implies. Secondly, we propose the Witness Simplicial VAE as an extension of the Simplicial Auto-Encoder to the variational setup using a Witness Complex for computing a simplicial regularization. The Witness Simplicial VAE is independent of any dimensionality reduction technique and seems to better preserve the persistent Betti numbers of a data set than a standard VAE, although it would still need some further improvements. Finally, the two first chapters of this master thesis can also be used as an introduction to Topological Data Analysis, General Topology and Computational Topology (or Algorithmic Topology), for any machine learning student, engineer or researcher interested in these areas with no background in topology. / Variations autokodare (VAE) är en av de mest kända djupa generativa modellerna. Efter att ha visat att standard VAE inte nödvändigtvis bevarar topologiska egenskaper, det vill säga formen på datan, mellan inmatningsdatan och det latenta rummet, försökte vi modifiera den så att topologin är bevarad. Det här skulle i synnerhet underlätta när man genomför interpolering i det latenta rummet. Denna avhandling består av två centrala bidrag. I första hand så utvecklar vi InvMap-VAE, som är en enkel metod att omvandla vilken metod inom dimensionalitetsreducering, givet dess inbäddning, till en generativ modell inom VAE ramverket, vilket ger en invers avbildning och dess tillhörande fördelar. För det andra så presenterar vi Witness Simplicial VAE som en förlängning av en Simplicial Auto-Encoder till dess variationella variant genom att använda ett vittneskomplex för att beräkna en simpliciel regularisering. Witness Simplicial VAE är oberoende av dimensionalitets reducerings teknik och verkar bättre bevara Betti-nummer av ett dataset än en vanlig VAE, även om det finns utrymme för förbättring. Slutligen så kan de första två kapitlena av detta examensarbete också användas som en introduktion till Topologisk Data Analys, Allmän Topologi och Beräkningstopologi (eller Algoritmisk Topologi) till vilken maskininlärnings student, ingenjör eller forskare som är intresserad av dessa ämnesområden men saknar bakgrund i topologi.