Gotowa bibliografia na temat „Reconstruction 3D pilotée par model”

3D Morphable models of the human body capture variations among subjects and are useful in reconstruction and editing applications. Current dental models use an explicit mesh scene representation and model only the teeth, ignoring the gum. In this work, we present the first parametric 3D morphable dental model for both teeth and gum. Our model uses an implicit scene representation and is learned from rigidly aligned scans. It is based on a component-wise representation for each tooth and the gum, together with a learnable latent code for each of such components. It also learns a template shape thus enabling several applications such as segmentation, interpolation and tooth replacement. Our reconstruction quality is on par with the most advanced global implicit representations while enabling novel applications. The code will be available at https://github.com/cong-yi/DMM

Recovery of articulated 3D structure from 2D observations is a challenging computer vision problem with many applications. Current learning-based approaches achieve state-of-the-art accuracy on public benchmarks but are restricted to specific types of objects and motions covered by the training datasets. Model-based approaches do not rely on training data but show lower accuracy on these datasets. In this paper, we introduce a model-based method called Structure from Articulated Motion (SfAM), which can recover multiple object and motion types without training on extensive data collections. At the same time, it performs on par with learning-based state-of-the-art approaches on public benchmarks and outperforms previous non-rigid structure from motion (NRSfM) methods. SfAM is built upon a general-purpose NRSfM technique while integrating a soft spatio-temporal constraint on the bone lengths. We use alternating optimization strategy to recover optimal geometry (i.e., bone proportions) together with 3D joint positions by enforcing the bone lengths consistency over a series of frames. SfAM is highly robust to noisy 2D annotations, generalizes to arbitrary objects and does not rely on training data, which is shown in extensive experiments on public benchmarks and real video sequences. We believe that it brings a new perspective on the domain of monocular 3D recovery of articulated structures, including human motion capture.

AbstractThe computational prediction of atomistic structure is a long-standing problem in physics, chemistry, materials, and biology. Conventionally, force-fields or ab initio methods determine structure through energy minimization, which is either approximate or computationally demanding. This accuracy/cost trade-off prohibits the generation of synthetic big data sets accounting for chemical space with atomistic detail. Exploiting implicit correlations among relaxed structures in training data sets, our machine learning model Graph-To-Structure (G2S) generalizes across compound space in order to infer interatomic distances for out-of-sample compounds, effectively enabling the direct reconstruction of coordinates, and thereby bypassing the conventional energy optimization task. The numerical evidence collected includes 3D coordinate predictions for organic molecules, transition states, and crystalline solids. G2S improves systematically with training set size, reaching mean absolute interatomic distance prediction errors of less than 0.2 Å for less than eight thousand training structures — on par or better than conventional structure generators. Applicability tests of G2S include successful predictions for systems which typically require manual intervention, improved initial guesses for subsequent conventional ab initio based relaxation, and input generation for subsequent use of structure based quantum machine learning models.

Abstract Funding Acknowledgements Type of funding sources: Other. Main funding source(s): National Research Agency (ANR) French Federation of Cardiology : “Aide à la recherche par équipe 2018, Cardiopathies de l’enfant” Introduction After percutaneous implantation of an atrial septal defect (ASD) occluder device, a complex healing process leads to the device coverage within several months. However, an unexplained incomplete device coverage is at risk of complications such as thrombosis or infectious endocarditis. Purpose The aim of the study was to assess the device coverage process of ASD occluder devices in a chronic sheep model using micro-CT technology. Methods After percutaneous creation of an ASD by catheterization, 8 ewes (mean age 5.4 ± 0.7 yo and mean weight 55.6 ± 7.9 kg) were implanted with a 16-mm Nit-Occlud ASD-R occluder (PFM medical, Cologne, Germany) and were followed for 1 month (N = 3) and 3 months (N =5). After heart explantation, a iodine contrast agent was used to enhance the tissue signal. The device coverage was then assessed by micro-CT and the results were compared to histology, used as the gold standard for healing evaluation. The micro-CT image resolution was 41.7 µm. Reconstruction was performed in 2D and 3D with Amira® software, allowing to obtain images that were exploited by a code to measure the surface for each disk of the analyzed devices. Histological study was performed after resin embedding and Richardson blue staining was used. The pathologist was blinded to the duration of animals’ follow-up and micro-CT results. Results ASD creation and device closure was successful in 100% animals without complications. Following heart explantation, macroscopic assessment of devices showed that the coverage was complete for the left-side disk regardless of the duration of the follow-up and variable for the right-side disk, depending of the protrusion of this disk. 2D and 3D micro-CT analysis allowed an accurate evaluation of device coverage of each disk and was overall well correlated to histology slices (cf Figure). Surface calculation from micro-CT images showed that the median surface of coverage was 93 ± 8% for the left-side disk and 55 ± 31% for the right-side disk. Conclusion This preliminary study made the proof of concept that micro-CT is a reliable tool to assess the coverage of intra-cardiac occluders in vitro. The translation to clinical practice is challenging but would allow an individual follow-up, to avoid thrombotic or infective complications. Abstract Figure.

Avec l’avancée des technologies de la microélectronique, l’architecture des composants électroniques devient de plus en plus compliquée. Or, la connaissance des caractéristiques dimensionnelles des structures réalisées est importante pour pouvoir comprendre et optimiser le comportement de ces composants. C’est pourquoi il existe un besoin de développer des méthodes de mesure tridimensionnelles rapides et non destructives.Le microscope électronique à balayage (SEM) est largement utilisé pour réaliser des mesures dimensionnelles car il répond aux problématiques de rapidité et de non-destructivité. Cependant l’obtention d’informations tridimensionnelles quantitatives et précises est un challenge.Grâce à un microscope électronique dont le faisceau électronique peut être incliné, il est possible d’obtenir des images à différents angles de vue. A partir de l’analyse de ces images, la hauteur et l’angle des flancs du motif observé peuvent être déterminés géométriquement.Cependant, l’imagerie électronique étant le résultat des interactions électrons-matière, il est important de comprendre l’origine de la formation des images SEM, pour pouvoir les analyser correctement. C’est pourquoi une étude a été menée grâce à un logiciel de simulation physique pour observer et comprendre l’impact de la topographie d’un motif sur l’image SEM résultante.A partir de ces observations, des métriques ont été créées sur les images SEM pour les analyser quantitativement. Un modèle linéaire a ensuite été créé grâce aux simulations physiques pour estimer les grandeurs topographiques à partir de ces métriques. Il a ensuite été calibré sur des mesures SEM réelles, en les comparant à des mesures tridimensionnelles de référence par microscopie à force atomique (AFM). Ce modèle a été créé pour la reconstruction de motifs de type "ligne" en silicium gravé. Grâce à ce modèle, des reconstructions de motifs réelles ont été réalisées. Enfin un travail sur la création d’un modèle pour les motifs de type "tranchée" et "dense" en silicium gravé a été initié
With the advancement of microelectronics technologies, the architecture of electronic components is becoming increasingly complicated. However, knowledge of the dimensional characteristics of the structures is important in order to be able to understand and optimize the behavior of these components. This is why there is a need to develop rapid, non-destructive three-dimensional measurement methods.The scanning electron microscope (SEM) is widely used to carry out dimensional measurements because it responds to the problems of speed and non-destructivity. However, obtaining quantitative and precise three-dimensional information is a challenge.Thanks to an electron microscope whose electron beam can be tilted, it is possible to obtain images at different viewing angles. From the analysis of these images, the height and the sidewall angles of the observed pattern can be determined geometrically.However, since electronic imaging is the result of electron-matter interactions, it is important to understand the origin of the formation of SEM images, in order to be able to analyze them correctly. This is why a study was carried out using physical simulation software to observe and understand the impact of the topography of a pattern on the resulting SEM image.From these observations, metrics were created on the SEM images to analyze them quantitatively.A linear model was then created using physical simulations to estimate the topographic quantities from these metrics. It was then calibrated on real SEM measurements, by comparing them to three-dimensional reference measurements by atomic force microscopy (AFM). This model was created for the reconstruction of “line” type patterns in etched silicon. Thanks to this model, reconstructions of real patterns were made. Finally, work was started on the creation of a model for "trench" and "dense" type patterns in etched silicon