公开(公告)号：US20240303962A1

公开(公告)日：2024-09-12

申请号：US18020914

申请日：2022-04-22

Applicant: BEIJING BAIDU NETCOM SCIENCE TECHNOLOGY CO., LTD.

Inventor： Bi LI ,  Nan PENG ,  Teng XI ,  Gang ZHANG

IPC: G06V10/50 ,  G06V10/26 ,  G06V10/42 ,  G06V10/44

CPC classification number: G06V10/50 ,  G06V10/26 ,  G06V10/42 ,  G06V10/44

Abstract: A method of determining an image feature, an electronic device, and a storage medium are provided, which relate to the field of artificial intelligence technology, in particular to fields of computer vision and depth learning technology, and may be applied to scenarios such as image processing and image recognition. The method includes: dividing an original image into a plurality of local images as an image to be processed, and each local image includes a plurality of image blocks; determining a local feature of the image to be processed according to a relationship between each image block in each local image; and determining a global feature of the image to be processed according to a relationship between a first image block at a preset position in a local image and one or more second image blocks at the preset position in other local images of the plurality of local images.

公开(公告)号：US20240265687A1

公开(公告)日：2024-08-08

申请号：US18020396

申请日：2022-04-22

Applicant: BEIJING BAIDU NETCOM SCIENCE TECHNOLOGY CO., LTD.

Inventor： Bi LI ,  Nan PENG ,  Teng XI ,  Gang ZHANG

IPC: G06V10/80 ,  G06V10/77

CPC classification number: G06V10/806 ,  G06V10/7715

Abstract: A method of fusing an image feature, an electronic device, and a storage medium are provided, which relate to the field of artificial intelligence, in particular to fields of computer vision and depth learning, and may be applied to scenarios such as image processing and image recognition. The method includes: inputting an image into a first image processing model among N serially connected image processing models, to obtain an output feature of the first image processing model, an i-th image processing model includes a first shared layer to an i-th shared layer, i=1, . . . , N, and N is a natural number greater than or equal to 2; inputting an output feature of a j-th image processing model into a (j+1)-th image processing model, to obtain an output feature of the (j+1)-th image processing model, j=1, . . . , N−1; and fusing the output features of the N image processing models.

公开(公告)号：US20230186607A1

公开(公告)日：2023-06-15

申请号：US18148174

申请日：2022-12-29

Applicant: Beijing Baidu Netcom Science Technology Co., Ltd.

Inventor： Nan PENG ,  Bi LI ,  Teng XI ,  Gang ZHANG

IPC: G06V10/771 ,  G06V10/774 ,  G06V10/72 ,  G06V10/77

CPC classification number: G06V10/771 ,  G06V10/774 ,  G06V10/72 ,  G06V10/7715

Abstract: A multi-task identification method, a training method, an electronic device, and a storage medium are provided, which relate to a field of an artificial intelligence technology, in particular to fields of deep learning, image processing and computer vision technologies, and may be applied to scenarios such as human faces. A specific implementation solution includes: obtaining first intermediate feature data according to an image to be identified; selecting a feature extraction strategy having a greatest matching degree with the image to be identified from a plurality of feature extraction strategies based on a target selection strategy and the first intermediate feature data, so as to obtain a target feature extraction strategy; processing the first intermediate feature data based on the target feature extraction strategy, to obtain second intermediate feature data; and obtaining a multi-task identification result for the image to be identified according to the second intermediate feature data.

公开(公告)号：US20230147202A1

公开(公告)日：2023-05-11

申请号：US18153292

申请日：2023-01-11

Applicant: BEIJING BAIDU NETCOM SCIENCE TECHNOLOGY CO., LTD.

Inventor： Bi LI ,  Nan PENG ,  Teng XI ,  Gang ZHANG

IPC: G06V40/16 ,  G06V10/80 ,  G06V10/75

CPC classification number: G06V40/172 ,  G06V40/171 ,  G06V10/806 ,  G06V10/751

Abstract: A method for face recognition is disclosed. The method includes: obtaining an image to be recognized; extracting an image feature of the image to be recognized; obtaining a fused feature corresponding to a reference image of reference image; determining a similarity between the image feature of the image to be recognized and the fused feature corresponding to the reference image of the reference images to obtain a determination result of the similarity; and determining, based on the obtained determination result of the similarity, a face recognition result of the image to be recognized. It can be learned that recognition precision of a human face having a shielding object and a human face not having the shielding object may be ensured by means of this solution.