Abstract: A computer-implemented method of machine-learning including obtaining a dataset of 3D point clouds. Each 3D point cloud includes at least one object. Each 3D point cloud is equipped with a specification of one or more graphical user-interactions each representing a respective selection operation of a same object in the 3D point cloud. The method further includes teaching, based on the dataset, a neural network configured for segmenting an input 3D point cloud including an object. The segmenting is based on the input 3D point cloud and on a specification of one or more input graphical user-interactions each representing a respective selection operation of the object in the 3D point cloud.

Abstract: The disclosure notably relates to a computer-implemented method for generating a 3D model representing a building. The method comprises providing a 2D floor plan representing a layout of the building. The method also comprises determining a semantic segmentation of the 2D floor plan. The method also comprises determining the 3D model based on the semantic segmentation. Such a method provides an improved solution for processing a 2D floor plan.

Abstract: A computer-implemented method of machine-learning is described that includes obtaining a dataset of virtual scenes. The dataset of virtual scenes belongs to a first domain. The method further includes obtaining a test dataset of real scenes. The test dataset belongs to a second domain. The method further includes determining a third domain. The third domain is closer to the second domain than the first domain in terms of data distributions. The method further includes learning a domain-adaptive neural network based on the third domain. The domain-adaptive neural network is a neural network configured for inference of spatially reconfigurable objects in a real scene. Such a method constitutes an improved method of machine learning with a dataset of scenes including spatially reconfigurable objects.

Abstract: A computer-implemented method of machine-learning. The method comprises providing a dataset of 3D modeled objects each representing a mechanical part. Each 3D modeled object comprises a specification of a geometry of the mechanical part. The method further comprises learning a set of parameterization vectors each respective to a respective 3D modeled object of the dataset and a neural network configured to take as input a parameterization vector and to output a representation of a 3D modeled object usable in a differentiable simulation-based shape optimization. The learning comprises minimizing a loss that penalizes, for each 3D modeled object of the dataset, a disparity between the output of the neural network for an input parameterization vector respective to the 3D modeled object and a representation of the 3D modeled object. The representation of the 3D modeled object is usable in a differentiable simulation-based shape optimization.

Abstract: A computer-implemented method for training a deep-learning generative model configured to output 3D modeled objects each representing a mechanical part or an assembly of mechanical parts. The method comprises obtaining a dataset of 3D modeled objects and training the deep-learning generative model based on the dataset. The training includes minimization of a loss. The loss includes a term that penalizes, for each output respective 3D modeled object, one or more functional scores of the respective 3D modeled object. Each functional score measures an extent of non-respect of a respective functional descriptor among one or more functional descriptors, by the mechanical part or the assembly of mechanical parts. This forms an improved solution with respect to outputting 3D modeled objects each representing a mechanical part or an assembly of mechanical parts.

Abstract: Described is a computer-implemented method for partitioning a 3D scene into a plurality of zones, each zone representing an area or a volume of the 3D scene and being processed by a computing resource. The method comprises obtaining a 3D scene comprising one or more objects, each object generating a computing resource cost, computing a first map that represents a density of computing costs of the provided 3D scene, defining a second map that represents constraints on the shapes of zones that will be obtained as a result of a partitioning of the 3D scene, discretizing the obtained 3D scene into cells by computing a space quantization of the 3D scene free of dynamic objects, computing, for each cell, a computing cost from the first map of the 3D scene, aggregating the cells into one or more zones in accordance with the second map.

Abstract: A computer-implemented method of machine-learning including obtaining a dataset of 3D point clouds. Each 3D point cloud includes at least one object. Each 3D point cloud is equipped with a specification of one or more graphical user-interactions each representing a respective selection operation of a same object in the 3D point cloud. The method further includes teaching, based on the dataset, a neural network configured for segmenting an input 3D point cloud including an object. The segmenting is based on the input 3D point cloud and on a specification of one or more input graphical user-interactions each representing a respective selection operation of the object in the 3D point cloud.

Abstract: A computer-implemented method of machine-learning is described that includes obtaining a dataset of virtual scenes. The dataset of virtual scenes belongs to a first domain. The method further includes obtaining a test dataset of real scenes. The test dataset belongs to a second domain. The method further includes determining a third domain. The third domain is closer to the second domain than the first domain in terms of data distributions. The method further includes learning a domain-adaptive neural network based on the third domain. The domain-adaptive neural network is a neural network configured for inference of spatially reconfigurable objects in a real scene. Such a method constitutes an improved method of machine learning with a dataset of scenes including spatially reconfigurable objects.

Abstract: A computer-implemented method of machine-learning is described that includes obtaining a test dataset of scenes. The test dataset belongs to a test domain. The method includes obtaining a domain-adaptive neural network. The domain-adaptive neural network is a machine-learned neural network taught using data obtained from a training domain. The domain-adaptive neural network is configured for inference of spatially reconfigurable objects in a scene of the test domain. The method further includes determining an intermediary domain. The intermediary domain is closer to the training domain than the test domain in terms of data distributions. The method further includes inferring, by applying the domain-adaptive neural network, a spatially reconfigurable object from a scene of the test domain transferred on the intermediary domain. Such a method constitutes an improved method of machine learning with a dataset of scenes comprising spatially reconfigurable objects.

Abstract: It is provided a computer-implemented method for performing body posture tracking, comprising the steps of collecting (S10) depth measurements of a body with at least one depth sensor; collecting (S20) inertial measurements with at least one inertial sensor attached to the body; and determining (S30) at least one posture of the body as a function of the depth measurements and the inertial measurements. Such a method improves the field of body posture tracking.

Abstract: The invention notably relates to a computer-implemented method for designing a 3D assembly of modeled objects. The method comprises rendering on a second computer a 3D assembly of modeled objects by merging a second 3D modeled object with at least one raster image of a first 3D modeled object, the at least one raster image having being streamed from a first computer to the second computer; sending from the second computer to the first computer first data related to the second 3D modeled object for contact computation between the first and second 3D modeled objects; and computing on the first computer a contact between the first and second 3D modeled objects.

Abstract: Described is a computer-implemented method for partitioning a 3D scene into a plurality of zones, each zone representing an area or a volume of the 3D scene and being processed by a computing resource. The method comprises obtaining a 3D scene comprising one or more objects, each object generating a computing resource cost, computing a first map that represents a density of computing costs of the provided 3D scene, defining a second map that represents constraints on the shapes of zones that will be obtained as a result of a partitioning of the 3D scene, discretizing the obtained 3D scene into cells by computing a space quantization of the 3D scene free of dynamic objects, computing, for each cell, a computing cost from the first map of the 3D scene, aggregating the cells into one or more zones in accordance with the second map.

Abstract: A computer implemented method for learning a function configured for captioning a region of an image. The method comprises providing a dataset of triplets each including a respective image, a respective region of the respective image, and a respective caption of the respective region. The method also comprises learning, with the dataset of triplets, a function that is configured to generate an output caption based on an input image and on an input region of the input image. Such a method constitutes an improved solution for captioning a region of an image.

Abstract: Described is a computer-implemented method for partitioning a 3D scene into a plurality of zones, each zone representing an area or a volume of the 3D scene and being processed by a computing resource. The method comprises obtaining a 3D scene comprising one or more objects, each object generating a computing resource cost, computing a first map that represents a density of computing costs of the provided 3D scene, defining a second map that represents constraints on the shapes of zones that will be obtained as a result of a partitioning of the 3D scene, discretizing the obtained 3D scene into cells by computing a space quantization of the 3D scene free of dynamic objects, computing, for each cell, a computing cost from the first map of the 3D scene, aggregating the cells into one or more zones in accordance with the second map.

Abstract: The invention is directed to a communication middleware for managing multicast channels between a server and a client connected through a communication network, wherein the communication middleware manages at least one compulsory multicast channel for delivering compulsory data to the client and the communication middleware manages at least one optional multicast channel for delivering optional data to the client.

Abstract: The invention notably relates to a computer-implemented method for simulating a 3D scene. The simulation is carried out with a set of computing resources running in parallel. The method comprises partitioning a 3D scene into a plurality of zones. Each zone is sized to satisfy real-time computing constraint by one computing resource of the set. The method comprises assigning each zone of the plurality to a computing resource, computing an estimation of a load of each computing resource and determining whether one or more computing resources are over-loaded or under-loaded, computing, for each zone, a contribution of the zone to the load of the computing resource to which the zone is assigned, reassigning one or more zones of a computing resource that is over-loaded or under-loaded to another computing resource, the reassignment resulting from the computed contributions of the zones with a combinatorial optimization algorithm.

Abstract: The invention notably relates to a memory storage having a linear track and having recorded thereon a multi-resolution image system of an object, the multi-resolution image system including a set of images, each image representing the object and having a respective resolution, wherein the recording is according to a continuous injection from a space-filling curve of the set of images to the linear track, the space-filling curve interlaces the different images, and the intersection between the space-filling curve and each image is on a Hilbert curve. The invention improves the way to record a multi-resolution image system of an object on a memory storage.

Abstract: The invention notably relates to a computer-implemented method for compressing data representing values of a physical attribute in a predetermined space. The method comprises providing a signal that includes a mapping from leaf cells of a hierarchical subdivision of the predetermined space each onto a respective coefficient representative of a value of the physical attribute at the respective leaf cell. The method also comprises determining a discrete wavelet transform of the signal and encoding the determined discrete wavelet transform. The method provides an improved way to compress a modeled object that represents a real object.

Abstract: The disclosure notably relates to a computer-implemented method for determining a function configured to determine a semantic segmentation of a 2D floor plan representing a layout of a building. The method comprises providing a dataset comprising 2D floor plans each associated to a respective semantic segmentation. The method also comprises learning the function based on the dataset. Such a method provides an improved solution for processing a 2D floor plan.

Abstract: The invention notably relates to a computer-implemented method for displaying a 3D assembly of modeled objects. The method comprises streaming from a first computer to a second computer at least one raster image of a first 3D modeled object, and rendering on the second computer the 3D assembly of modeled objects by merging a second 3D modeled object with the streamed at least one raster image.