Dataset for learning a function taking images as inputs

    公开(公告)号:US10984317B2

    公开(公告)日:2021-04-20

    申请号:US16011359

    申请日:2018-06-18

    Abstract: The disclosure notably relates to a computer-implemented method for forming a dataset. The dataset is configured for learning a function. The function takes as inputs images of instances of one or more classes of real objects. The method comprises, for each class, providing a parametric model of the class, generating a plurality of 3D modeled objects with the parametric model, and adding to the dataset, for each one of the plurality of 3D modeled objects, one or more corresponding images. The generating includes traversing one or more of the set of ranges. The method constitutes an improved solution for forming a dataset configured for learning a function.

    3D object localization with quantization

    公开(公告)号:US10235815B2

    公开(公告)日:2019-03-19

    申请号:US15388830

    申请日:2016-12-22

    Inventor: Jonathan Cloux

    Abstract: The invention notably relates to a computer-implemented method for localizing a 3D modeled object in a 3D scene, the method comprising a positioning of the 3D modeled object in the 3D scene, the positioning being performed following an algorithm that rewards, for each of first couples made of two 3D points of the 3D modeled object and their respective associated normal vectors, a match with a respective second couple made of two 3D point of the 3D scene and its respective associated normal vectors, the first couple being positioned substantially on the second couple, the match between the first couple and the second couple amounting to a substantial integer equality between the value of a quantized descriptor for the first couple and the value of the quantized descriptor for the second couple. This improves the localizing a 3D modeled object in a 3D scene.

    DATASET FOR LEARNING A FUNCTION TAKING IMAGES AS INPUTS

    公开(公告)号:US20180365565A1

    公开(公告)日:2018-12-20

    申请号:US16011359

    申请日:2018-06-18

    Abstract: The disclosure notably relates to a computer-implemented method for forming a dataset. The dataset is configured for learning a function. The function takes as inputs images of instances of one or more classes of real objects. The method comprises, for each class, providing a parametric model of the class, generating a plurality of 3D modeled objects with the parametric model, and adding to the dataset, for each one of the plurality of 3D modeled objects, one or more corresponding images. The generating includes traversing one or more of the set of ranges. The method constitutes an improved solution for forming a dataset configured for learning a function.

    3D object localization with descriptor

    公开(公告)号:US10062217B2

    公开(公告)日:2018-08-28

    申请号:US15389572

    申请日:2016-12-23

    Inventor: Jonathan Cloux

    Abstract: The invention notably relates to a computer-implemented method for localizing a 3D modeled object in a 3D scene, the method comprising a positioning of the 3D modeled object in the 3D scene, the positioning being performed following an algorithm that rewards, for each of first couples made of two 3D points of the 3D modeled object and their respective associated normal vectors, a match with a respective second couple made of two 3D point of the 3D scene and its respective associated normal vectors, the match between the first couple and the second couple amounting to a substantial equality between the value of a descriptor for the first couple and the value of the descriptor for the second couple, the descriptor being variable under mirror symmetries. This improves the localizing a 3D modeled object in a 3D scene.

    Designing a 3D modeled object with 2D views

    公开(公告)号:US09886528B2

    公开(公告)日:2018-02-06

    申请号:US14290639

    申请日:2014-05-29

    CPC classification number: G06F17/50 G06T17/00

    Abstract: It is provided a computer-implemented method designing a three-dimensional modeled object, comprising the steps of providing (S10) a plurality of two-dimensional views of the modeled object, a three-dimensional wireframe graph, associating (S20), to each vertex of the wireframe graph, a local radial order between all the edges incident to the vertex, and then determining (S30) edge cycles, by browsing the wireframe graph following the local radial orders associated to the vertices. Such a method improves the design of a 3D modeled object.

Patent Agency Ranking