公开(公告)号：US20230186467A1

公开(公告)日：2023-06-15

申请号：US18049205

申请日：2022-10-24

申请人： OWKIN, INC. ,  OWKIN FRANCE SAS

发明人： Pierre Courtiol ,  Eric W. Tramel ,  Marc Sanselme ,  Gilles Wainrib

IPC分类号： G06T7/00 ,  G06V20/69 ,  G06F18/21 ,  G06V10/764 ,  G06V10/82

CPC分类号： G06T7/0012 ,  G06V20/698 ,  G06V20/695 ,  G06F18/2163 ,  G06V10/764 ,  G06V10/82 ,  G06T2207/20084 ,  G06T2207/30024

摘要： A method and apparatus of a device that classifies an image is described. In an exemplary embodiment, the method includes tiling at least one region of interest of the input image into a set of tiles. For each tile, the method includes extracting a feature vector of the tile by applying a convolutional neural network, wherein a feature is a local descriptor of the tile; and computing a score of the tile from the extracted feature vector, said tile score being representative of a contribution of the tile into a classification of the input image. The method also includes sorting a set of the tile scores and selecting a subset of the tile scores based on their value and/or their rank in the sorted set. The method also includes applying a classifier to the selected tile scores in order to classify the input image.

公开(公告)号：US20200250398A1

公开(公告)日：2020-08-06

申请号：US16778179

申请日：2020-01-31

申请人： Owkin Inc. ,  Owkin France SAS

发明人： Pierre Courtiol ,  Eric W. Tramel ,  Marc Sanselme ,  Gilles Wainrib

IPC分类号： G06K9/00 ,  G06K9/62 ,  G06K9/46 ,  G06T7/00

摘要： A method and apparatus of a device that classifies an image is described. In an exemplary embodiment, the method includes tiling at least one region of interest of the input image into a set of tiles. For each tile, the method includes extracting a feature vector of the tile by applying a convolutional neural network, wherein a feature is a local descriptor of the tile; and computing a score of the tile from the extracted feature vector, said tile score being representative of a contribution of the tile into a classification of the input image. The method also includes sorting a set of the tile scores and selecting a subset of the tile scores based on their value and/or their rank in the sorted set. The method also includes applying a classifier to the selected tile scores in order to classify the input image

公开(公告)号：US20210200989A1

公开(公告)日：2021-07-01

申请号：US17183321

申请日：2021-02-23

申请人： Owkin Inc. ,  Owkin France SAS

发明人： Pierre Courtiol ,  Eric W. Tramel ,  Marc Sanselme ,  Gilles Wainrib

IPC分类号： G06K9/00 ,  G06K9/62 ,  G06K9/46 ,  G06T7/00

摘要： A method and apparatus of a device that classifies an image is described. In an exemplary embodiment, the method includes tiling at least one region of interest of the input image into a set of tiles. For each tile, the method includes extracting a feature vector of the tile by applying a convolutional neural network, wherein a feature is a local descriptor of the tile; and computing a score of the tile from the extracted feature vector, said tile score being representative of a contribution of the tile into a classification of the input image. The method also includes sorting a set of the tile scores and selecting a subset of the tile scores based on their value and/or their rank in the sorted set. The method also includes applying a classifier to the selected tile scores in order to classify the input image.

公开(公告)号：US12118715B2

公开(公告)日：2024-10-15

申请号：US16778179

申请日：2020-01-31

申请人： Owkin Inc. ,  Owkin France SAS

发明人： Pierre Courtiol ,  Eric W. Tramel ,  Marc Sanselme ,  Gilles Wainrib

IPC分类号： G06K9/00 ,  G06F18/21 ,  G06T7/00 ,  G06V10/764 ,  G06V10/82 ,  G06V20/69

CPC分类号： G06T7/0012 ,  G06F18/2163 ,  G06V10/764 ,  G06V10/82 ,  G06V20/695 ,  G06V20/698 ,  G06T2207/20084 ,  G06T2207/30024

摘要： A method and apparatus of a device that classifies an image is described. In an exemplary embodiment, the method includes tiling at least one region of interest of the input image into a set of tiles. For each tile, the method includes extracting a feature vector of the tile by applying a convolutional neural network, wherein a feature is a local descriptor of the tile; and computing a score of the tile from the extracted feature vector, said tile score being representative of a contribution of the tile into a classification of the input image. The method also includes sorting a set of the tile scores and selecting a subset of the tile scores based on their value and/or their rank in the sorted set. The method also includes applying a classifier to the selected tile scores in order to classify the input image.

公开(公告)号：US20240346651A1

公开(公告)日：2024-10-17

申请号：US18751788

申请日：2024-06-24

申请人： OWKIN, INC. ,  OWKIN FRANCE SAS

发明人： Pierre Courtiol ,  Eric W. Tramel ,  Marc Sanselme ,  Gilles Wainrib

IPC分类号： G06T7/00 ,  G06F18/21 ,  G06V10/764 ,  G06V10/82 ,  G06V20/69

CPC分类号： G06T7/0012 ,  G06F18/2163 ,  G06V10/764 ,  G06V10/82 ,  G06V20/695 ,  G06V20/698 ,  G06T2207/20084 ,  G06T2207/30024

摘要： A method and apparatus of a device that classifies an image is described. In an exemplary embodiment, the method includes tiling at least one region of interest of the input image into a set of tiles. For each tile, the method includes extracting a feature vector of the tile by applying a convolutional neural network, wherein a feature is a local descriptor of the tile; and computing a score of the tile from the extracted feature vector, said tile score being representative of a contribution of the tile into a classification of the input image. The method also includes sorting a set of the tile scores and selecting a subset of the tile scores based on their value and/or their rank in the sorted set. The method also includes applying a classifier to the selected tile scores in order to classify the input image.

公开(公告)号：US11482022B2

公开(公告)日：2022-10-25

申请号：US17183321

申请日：2021-02-23

申请人： Owkin Inc. ,  Owkin France SAS

发明人： Pierre Courtiol ,  Eric W. Tramel ,  Marc Sanselme ,  Gilles Wainrib

IPC分类号： G06K9/00 ,  G06V20/69 ,  G06K9/62 ,  G06T7/00 ,  G06V10/50

摘要： A method and apparatus of a device that classifies an image is described. In an exemplary embodiment, the method includes tiling at least one region of interest of the input image into a set of tiles. For each tile, the method includes extracting a feature vector of the tile by applying a convolutional neural network, wherein a feature is a local descriptor of the tile; and computing a score of the tile from the extracted feature vector, said tile score being representative of a contribution of the tile into a classification of the input image. The method also includes sorting a set of the tile scores and selecting a subset of the tile scores based on their value and/or their rank in the sorted set. The method also includes applying a classifier to the selected tile scores in order to classify the input image.