公开(公告)号：US20210192194A1

公开(公告)日：2021-06-24

申请号：US17022219

申请日：2020-09-16

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

Inventor： Zhizhen Chi ,  Fu Li ,  Hao Sun ,  Dongliang He ,  Xiang Long ,  Zhichao Zhou ,  Ping Wang ,  Shilei Wen ,  Errui Ding

IPC: G06K9/00 ,  G06F3/01 ,  G06N3/02

Abstract: The present application discloses a video-based human behavior recognition method, apparatus, device and storage medium, and relates to the technical field of human recognitions. The specific implementation scheme lies in: acquiring a human rectangle of each video frame of the video to be recognized, where each human rectangle includes a plurality of human key points, and each of the human key points has a key point feature; constructing a feature matrix according to the human rectangle of the each video frame; convolving the feature matrix with respect to a video frame quantity dimension to obtain a first convolution result and convolving the feature matrix with respect to a key point quantity dimension to obtain a second convolution result; inputting the first convolution result and the second convolution result into a preset classification model to obtain a human behavior category of the video to be recognized.

公开(公告)号：US11430265B2

公开(公告)日：2022-08-30

申请号：US17022219

申请日：2020-09-16

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

Inventor： Zhizhen Chi ,  Fu Li ,  Hao Sun ,  Dongliang He ,  Xiang Long ,  Zhichao Zhou ,  Ping Wang ,  Shilei Wen ,  Errui Ding

IPC: G06V40/20 ,  G06F3/01 ,  G06N3/02 ,  G06V20/40

Abstract: The present application discloses a video-based human behavior recognition method, apparatus, device and storage medium, and relates to the technical field of human recognitions. The specific implementation scheme lies in: acquiring a human rectangle of each video frame of the video to be recognized, where each human rectangle includes a plurality of human key points, and each of the human key points has a key point feature; constructing a feature matrix according to the human rectangle of the each video frame; convolving the feature matrix with respect to a video frame quantity dimension to obtain a first convolution result and convolving the feature matrix with respect to a key point quantity dimension to obtain a second convolution result; inputting the first convolution result and the second convolution result into a preset classification model to obtain a human behavior category of the video to be recognized.