METHOD AND APPARATUS WITH ATTENTION-BASED OBJECT ANALYSIS

    公开(公告)号:US20240153130A1

    公开(公告)日:2024-05-09

    申请号:US18342892

    申请日:2023-06-28

    CPC classification number: G06T7/73 G06T7/50 G06V10/25 G06V10/764 G06V10/771

    Abstract: An apparatus includes one or more processors configured to generate a plurality of feature maps having respective different resolutions based on an input image; and update, for each of the plurality of transformer layers, respective position estimation information comprising first position information of a respective bounding box corresponding to one object query and second position information of respective key points corresponding to the one object query, wherein each of the plurality of transformer layers includes a self-attention model configured to generate respective intermediate data by performing self-attention on respective content information on a feature of the input image; and a cross-attention model configured to generate respective output data by performing cross-attention on respective one or more feature maps among the plurality of feature maps and the respective generated intermediate data.

    METHOD AND APPARATUS WITH OBJECT DETECTION
    47.
    发明公开

    公开(公告)号:US20230186586A1

    公开(公告)日:2023-06-15

    申请号:US18078311

    申请日:2022-12-09

    CPC classification number: G06V10/20 G06N3/08 G06V10/82

    Abstract: An electronic device generates a feature map from an input image to perform object detection, classifies one or more objects included in the input image and determines one or more object regions including the one or more objects based on the feature map, classifies an ROI included in at least a portion of the objects and determines the ROI included in the input image based on the feature map, displays on the input image an indicator identifying a first object region of a first object where the ROI is determined and a feature point of a first ROI of the first object, and displays on the input image an indicator identifying a second object region of a second object where the ROI is not determined and a feature point of the second object region, to perform post-processing differently according to whether an ROI is determined in an object.

    METHOD AND APPARATUS WITH SELF-ATTENTION-BASED IMAGE RECOGNITION

    公开(公告)号:US20230154171A1

    公开(公告)日:2023-05-18

    申请号:US17720681

    申请日:2022-04-14

    CPC classification number: G06V10/82 G06N3/08 G06V10/40

    Abstract: A method with self-attention includes: obtaining a three-dimensional (3D) feature map; generating 3D query data and 3D key data by performing a convolution operation based on the 3D feature map; generating two-dimensional (2D) vertical data based on a vertical projection of the 3D query data and the 3D key data; generating 2D horizontal data based on a horizontal projection of the 3D query data and the 3D key data; determining an intermediate attention result through a multiplication based on the 2D vertical data and the 2D horizontal data; and determining a final attention result through a multiplication based on the intermediate attention result and the 3D feature map.

Patent Agency Ranking