Abstract: The disclosure notably relates to computer-implemented method for learning a neural network configured for inference, from a freehand drawing representing a 3D shape, of a solid CAD feature representing the 3D shape. The method includes providing a dataset including freehand drawings each representing a respective 3D shape, and learning the neural network based on the dataset. The method forms an improved solution for inference, from a freehand drawing representing a 3D shape, of a 3D modeled object representing the 3D shape.

Abstract: A method for processing a radiological image, in digital format includes at least one radiological anomaly detected using a convolutional neural network that has been trained to detect radiological anomalies on radiological examinations, the radiological image being characterized by the intensity of each of its pixels, and by at least one radiological anomaly influence map that assigns, for each pixel of the radiological image, a value representative of the proportion to which the pixel had an influence on the detection result of the radiological anomaly, which method is computer-implemented and comprises the steps of: normalizing the radiological anomaly influence maps to give normalized radiological anomaly influence maps; fusing the normalized radiological anomaly influence maps to give a single fused influence map; carrying out improvement processing on the image, using an intensity histogram, wherein the contribution of each pixel in the computing of the intensity histogram is weighted by the fused influ

Abstract: A computer-implemented method for forming a dataset configured for learning a neural network. The neural network is configured for inference, from a freehand drawing representing a 3D shape, of a solid CAD feature representing the 3D shape. The method includes generating one or more solid CAD feature includes each representing a respective 3D shape. The method also includes, for each solid CAD feature, determining one or more respective freehand drawings each representing the respective 3D shape, and inserting in the dataset, one or more training samples. Each training sample includes the solid CAD feature and a respective freehand drawing. The method forms an improved solution for inference, from a freehand drawing representing a 3D shape, of a 3D modeled object representing the 3D shape.

Abstract: The disclosure notably relates to a computer-implemented method for learning a neural network configured for inference, from a discrete geometrical representation of a 3D shape, of an editable feature tree representing the 3D shape. The editable feature tree includes a tree arrangement of geometrical operations applied to leaf geometrical shapes. The method includes obtaining a dataset including discrete geometrical representations each of a respective 3D shape, and obtaining a candidate set of leaf geometrical shapes. The method also includes learning the neural network based on the dataset and on the candidate set. The candidate set includes at least one continuous subset of leaf geometrical shapes. The method forms an improved solution for digitization.

Abstract: The disclosure notably relates to a computer-implemented method for forming a dataset configured for learning a neural network. The neural network is configured for inference, from a discrete geometrical representation of a 3D shape, of an editable feature tree representing the 3D shape. The editable feature tree comprises a tree arrangement of geometrical operations applied to leaf geometrical shapes. The method includes obtaining respective data pieces, and inserting a part of the data pieces in the dataset each as a respective training sample. The respective 3D shape of each of one or more first data pieces inserted in the dataset is identical to the respective 3D shape of respective one or more second data pieces not inserted in the dataset. The method forms an improved solution for digitization.

Abstract: The disclosure notably relates to a computer-implemented method for generating a structured three-dimensional (3D) model from a mesh. The method includes obtaining a mesh that comprises faces, each face of the mesh including a normal and principal curvature values; computing a distribution of the principal curvatures values over the whole mesh by counting the number of occurrences of discretized curvature values; identifying in the computed distribution one or more dominant ranges of principal curvature values; for each identified dominant range, computing one or more regions of the mesh that includes faces belonging to the identified dominant range; for each computed region, detecting a primitive type by using the curvatures values of all faces of the region and identifying parameters of the detected primitive by using the mesh surface of the region.

Abstract: The disclosure notably relates to a computer-implemented method for extracting a feature tree from a mesh. The method includes providing a mesh, computing a geometric and adjacency graph of the provided mesh, wherein each node of the graph represents one region of the mesh and comprises a primitive type and parameters of the region, each connection between two nodes is an intersection between the respective surfaces of the regions represented by the two connected nodes. The method also includes instantiating for each node of the graph, a surface based on the identified primitive type and parameters of the region.

Abstract: The disclosure notably relates to a computer-implemented method for 3D reconstruction. The method comprises providing a 3D point cloud representing a real object. The method also comprises fitting the 3D point cloud with parametric surfaces. The method also comprises defining a partition of the parametric surfaces into oriented facets which respect intersections between the parametric surfaces. The method also comprises determining, among the oriented facets of the partition, a set of facets that represents a skin of the real object. The determining comprises minimizing an energy. The energy includes a data term and a constraint term. The data term increasingly penalizes discarding facets, as a level of fit between a discarded facet and the 3D point cloud increases. The constraint term penalizes formation of non-skin geometry by kept facets. Such a method provides an improved solution for 3D reconstruction.

Abstract: The disclosure notably relates to a computer-implemented method for forming a dataset configured for learning a neural network. The neural network is configured for inference, from a discrete geometrical representation of a 3D shape, of an editable feature tree representing the 3D shape. The editable feature tree comprises a tree arrangement of geometrical operations applied to leaf geometrical shapes. The method includes obtaining respective data pieces, and inserting a part of the data pieces in the dataset each as a respective training sample. The respective 3D shape of each of one or more first data pieces inserted in the dataset is identical to the respective 3D shape of respective one or more second data pieces not inserted in the dataset. The method forms an improved solution for digitization.

Abstract: The disclosure notably relates to a computer-implemented method for extracting a feature tree from a mesh. The method includes providing a mesh, computing a geometric and adjacency graph of the provided mesh, wherein each node of the graph represents one region of the mesh and comprises a primitive type and parameters of the region, each connection between two nodes is an intersection between the respective surfaces of the regions represented by the two connected nodes. The method also includes instantiating for each node of the graph, a surface based on the identified primitive type and parameters of the region.

Abstract: The disclosure notably relates to computer-implemented method for learning a neural network configured for inference, from a freehand drawing representing a 3D shape, of a solid CAD feature representing the 3D shape. The method includes providing a dataset including freehand drawings each representing a respective 3D shape, and learning the neural network based on the dataset. The method forms an improved solution for inference, from a freehand drawing representing a 3D shape, of a 3D modeled object representing the 3D shape.

Abstract: The disclosure notably relates to a computer-implemented method for generating a structured three-dimensional (3D) model from a mesh. The method includes obtaining a mesh that comprises faces, each face of the mesh including a normal and principal curvature values; computing a distribution of the principal curvatures values over the whole mesh by counting the number of occurrences of discretized curvature values; identifying in the computed distribution one or more dominant ranges of principal curvature values; for each identified dominant range, computing one or more regions of the mesh that includes faces belonging to the identified dominant range; for each computed region, detecting a primitive type by using the curvatures values of all faces of the region and identifying parameters of the detected primitive by using the mesh surface of the region.

Abstract: The disclosure notably relates to a computer-implemented method for forming a dataset configured for learning a neural network. The neural network is configured for inference, from a freehand drawing representing a 3D shape, of a solid CAD feature representing the 3D shape. The method includes generating one or more solid CAD feature includes each representing a respective 3D shape. The method also includes, for each solid CAD feature, determining one or more respective freehand drawings each representing the respective 3D shape, and inserting in the dataset, one or more training samples. Each training sample includes the solid CAD feature and a respective freehand drawing. The method forms an improved solution for inference, from a freehand drawing representing a 3D shape, of a 3D modeled object representing the 3D shape.

Abstract: The disclosure notably relates to a computer-implemented method for learning a neural network configured for inference, from a discrete geometrical representation of a 3D shape, of an editable feature tree representing the 3D shape. The editable feature tree includes a tree arrangement of geometrical operations applied to leaf geometrical shapes. The method includes obtaining a dataset including discrete geometrical representations each of a respective 3D shape, and obtaining a candidate set of leaf geometrical shapes. The method also includes learning the neural network based on the dataset and on the candidate set. The candidate set includes at least one continuous subset of leaf geometrical shapes. The method forms an improved solution for digitization.

Abstract: The disclosure notably relates to a computer-implemented method for 3D reconstruction. The method comprises providing a 3D point cloud representing a real object. The method also comprises fitting the 3D point cloud with parametric surfaces. The method also comprises defining a partition of the parametric surfaces into oriented facets which respect intersections between the parametric surfaces. The method also comprises determining, among the oriented facets of the partition, a set of facets that represents a skin of the real object. The determining comprises minimizing an energy. The energy includes a data term and a constraint term. The data term increasingly penalizes discarding facets, as a level of fit between a discarded facet and the 3D point cloud increases. The constraint term penalizes formation of non-skin geometry by kept facets. Such a method provides an improved solution for 3D reconstruction.