Alexandre Meyer

Intuitive crowd behavior in dense urban environments using local laws

In games, entertainment, medical and architectural applications, the creation of populated virtual city environments has recently become widespread. In this paper we want to provide a technique that allows the simulation of up to 10,000 pedestrians walking in real-time. Simulation for such environments is difficult as a trade off needs to be found between realism and real-time simulation. This paper presents a pedestrian crowd simulation method aiming at improving the local and global reactions of the pedestrians. The method uses a subdivision of space into a 2D (two-dimensional) grid for pedestrian-to-pedestrian collision avoidance, while assigning goals to pedestrians to make their trajectories smoother and coherent. Goals are computed automatically and connected into a graph that reflects the structure of the city and triggers a spatial repartition of the density of pedestrians. In order to create realistic reactions when areas become crowded, local directions are stored and updated in real-time, allowing the apparition of pedestrian streams. Combining the different methods contributes to a more realistic model, while keeping a real-time frame rate for up to 10,000 simulated pedestrians.

Interactive Volumetric Textures

This paper presents a method for interactively rendering complex repetitive scenes such as landscapes, fur, organic tissues, etc. It is an adaptation to Z-buffer of volumetric textures, a ray-traced method, in order to use the power of existing graphics hardware. Our approach consists in slicing a piece of 3D geometry (one repetitive detail of the complex data) into a series of thin layers. A layer is a rectangle containing the shaded geometry that falls in that slice. These layers are used as transparent textures, that are mapped onto the underlying surface (e.g. a hill or an animal skin) with an extrusion offset. We show some results obtained with our first implementation, such as a scene of 13 millions of virtual polygons animated at 2.5 frames per second on a SGI O2.

Interactive Rendering of Trees with Shading and Shadows

The goal of this paper is the interactive rendering of 3D trees covering a landscape, with shading and shadows consistent with the lighting conditions. We propose a new IBR representation, consisting of a hierarchy of Bidirectional Textures, which resemble 6D lightfields. A hierarchy of visibility cube-maps is associated to this representation to improve the performance of shadow calculations. An example of hierarchy for a given tree can be a small branch plus its leaves (or needles), a larger branch, and the entire tree. A Bidirectional Texture (BT) provides a billboard image of a shaded object for each pair of view and light directions. We associate a BT for each level of the hierarchy. When rendering, the appropriate level of detail is selected depending on the distance of the tree from the viewpoint. The illumination reaching each level is evaluated using a visibility cube-map. Thus, we very efficiently obtain the shaded rendering of a tree with shadows without loosing details, contrary to mesh simplification methods. We achieved 7 to 20 fps fly-throughs of a scene with 1000 trees.

Framework for reliable, real-time facial expression recognition for low resolution images

Automatic recognition of facial expressions is a challenging problem specially for low spatial resolution facial images. It has many potential applications in human-computer interactions, social robots, deceit detection, interactive video and behavior monitoring. In this study we present a novel framework that can recognize facial expressions very efficiently and with high accuracy even for very low resolution facial images. The proposed framework is memory and time efficient as it extracts texture features in a pyramidal fashion only from the perceptual salient regions of the face. We tested the framework on different databases, which includes Cohn-Kanade (CK+) posed facial expression database, spontaneous expressions of MMI facial expression database and FG-NET facial expressions and emotions database (FEED) and obtained very good results. Moreover, our proposed framework exceeds state-of-the-art methods for expression recognition on low resolution images.

Point-based rendering of trees

The goal of this paper is the interactive and realistic rendering of 3D trees covering a landscape. The landscape is composed by instantiating one or more block of vegetation on the terrain. A block of vegetation is composed by a single or a compact group of trees. For these blocks of vegetation, we propose a new representation based on triangle+point primitives organized into a regular spatial structure (grid). This structure is de ned onto easily adapt the level of details (LOD) of each subpart (cell) of the vegetation element. During the rendering process, we determine a global level of details for each block of vegetation. Then, we re ne it for each cell according to the following heuristic: leaves or branches on the rear of tree or inside the forest are statistically less visible than front ones and then can be rendered coarsely. As a result, our method greatly decrease the number of rendered primitives by preserving realism. This allows rendering of large landscape in interactive time, for a camera far away until inside.

Human vision inspired framework for facial expressions recognition

We present a novel human vision inspired framework that can recognize facial expressions very efficiently and accurately. We propose to computationally process small, salient region of the face to extract features as it happens in human vision. To determine which facial region(s) is perceptually salient for a particular expression, we conducted a psycho-visual experimental study with an eye-tracker. A novel feature space conducive for recognition task is proposed, which is created by extracting Pyramid Histogram of Orientation Gradients features only from the salient facial regions. By processing only salient regions, proposed framework achieved two goals: (a) reduction in computational time for feature extraction (b) reduction in feature vector dimensionality. The proposed framework achieved automatic expression recognition accuracy of 95.3% on extended Cohn-Kanade (CK+) facial expression database for six universal facial expressions.

Feature points based facial animation retargeting

We present a method for transferring facial animation in real-time. The source animation may be an existing 3D animation or 2D data providing by a video tracker or a motion capture system. Based on two sets of feature points manually selected on the source and target faces (the only manual work required), a RBF network is trained and provides a geometric transformation between the two faces. At each frame, the RBF transformation is applied on the new feature points positions of the source face, resulting in new positions for target feature points according with the expression of the source face and the morphology of the target face. According to their displacements along time, we deform the target mesh on the GPU with the linear blend skinning (LBS) method. In order to make our approach attractive to novice user, we propose a procedural technique to automatically rig the target face by generating vertices weights for the skinning deformation. To summarize, our method provides interactive expression transfer with a minimal human intervention during setup and accepts various kind of animation sources.

Pain detection through shape and appearance features

In this paper we are proposing a novel computer vision system that can recognize expression of pain in videos by analyzing facial features. Usually pain is reported and recorded manually and thus carry lot of subjectivity. Manual monitoring of pain makes difficult for the medical practitioners to respond quickly in critical situations. Thus, it is desirable to design such a system that can automate this task. With our proposed model pain monitoring can be done in real-time without any human intervention. We propose to extract shape information using pyramid histogram of orientation gradients (PHOG) and appearance information using pyramid local binary pattern (PLBP) in order to get discriminative representation of face. We tested our proposed model on UNBC-McMaster Shoulder Pain Expression Archive Database and recorded results that exceeds state-of-the-art.

Exploring human visual system: Study to aid the development of automatic facial expression recognition framework

This paper focus on understanding human visual system when it decodes or recognizes facial expressions. Results presented can be exploited by the computer vision research community for the development of robust descriptor based on human visual system for facial expressions recognition. We have conducted psycho-visual experimental study to find which facial region is perceptually more attractive or salient for a particular expression. Eye movements of 15 observers were recorded with an eye-tracker in free viewing conditions as they watch a collection of 54 videos selected from Cohn-Kanade facial expression database, showing six universal facial expressions. The results of the study shows that for some facial expressions only one facial region is perceptually more attractive than others. Other cases shows the attractiveness of two to three facial regions. This paper also proposes a novel framework for automatic recognition of expressions which is based on psycho-visual study.

User-guided shape from shading to reconstruct fine details from a single photograph

Many real objects, such as faces, sculptures, or low-reliefs are composed of many detailed parts that can not be easily modeled by an artist nor by 3D scanning. In this paper, we propose a new shape from shading (SfS) approach to rapidly model details of these objects such as wrinkles and reliefs of surfaces from one photograph. The method first determines the surfaces flat areas in the photograph. Then, it constructs a graph of relative altitudes between each of these flat areas. We circumvent the ill-posed problem of shape from shading by having the user set if some of these flat areas are a local maximum or a local minimum; additional points can be added by the user (e.g. at discontinuous creases) - this is the only user input. We use an intuitive mass-spring based minimization to determine the final position of these flat areas and a fast-marching method to generate the surface. This process can be iterated until the user is satisfied with the resulting surface. We illustrate our approach on real faces and low-relief photographs.