US 11,816,930 B2
Behavior recognition method and apparatus, medium, and electronic device
Yunhui Zhang, Shenzhen (CN); Zhikai Zhang, Shenzhen (CN); and Cong Yang, Shenzhen (CN)
Assigned to Shenzhen Horizon Robotics Technology Co., Ltd.
Filed by Shenzhen Horizon Robotics Technology Co., Ltd., Nanshan Dist (CN)
Filed on Dec. 6, 2021, as Appl. No. 17/543,680.
Claims priority of application No. 202011467345.X (CN), filed on Dec. 14, 2020.
Prior Publication US 2022/0188537 A1, Jun. 16, 2022
Int. Cl. G06V 40/20 (2022.01); G06V 20/40 (2022.01)
CPC G06V 40/20 (2022.01) [G06V 20/46 (2022.01); G06V 20/44 (2022.01)] 14 Claims
OG exemplary drawing
 
1. A behavior recognition method, including:
obtaining a sequence of image blocks containing a target part of a target object from a plurality of video frames obtained by a camera apparatus disposed in a mobile device;
performing feature vector extraction that is based on a first preset behavior to the sequence of image blocks, to obtain a first feature vector corresponding to the sequence of image blocks;
determining a behavior type to which a behavior of the target object belongs based on the first feature vector; and
if the behavior type to which the behavior of the target object belongs satisfies a first behavior type, determining the behavior of the target object based on a result of performing behavior recognition on a second preset behavior for the sequence of image block;
wherein the determining a behavior type to which a behavior of the target object belongs based on the first feature vector includes:
for any image block in the sequence of image blocks, determining, based on the first feature vector of the image block, confidences that the behavior of the target object is each of a plurality of first preset behaviors, to obtain a plurality of first confidences; and
determining the behavior type to which the behavior of the target object belongs based on the plurality of first confidences respectively corresponding to each image block in the sequence of image blocks and first thresholds corresponding to the plurality of first preset behaviors;
wherein at least one first preset behavior in the plurality of first preset behaviors belongs to the first behavior type, and at least one first preset behavior in the plurality of first preset behaviors belongs to a second behavior type and
wherein the determining the behavior of the target object based on a result of performing behavior recognition on a second preset behavior for the sequence of image blocks includes:
performing feature vector extraction that is based on the second preset behavior to the sequence of image blocks, to obtain a second feature vector corresponding to the sequence of image blocks;
for any image block in the sequence of image blocks, determining, based on the second feature vector of the image block, confidences that the behavior of the target object is each of at least two second preset behaviors, to obtain at least two second confidences; and
the behavior of the target object based on the at least two second confidences respectively corresponding to each image block in the sequence of image blocks and second thresholds respectively corresponding to the at least two second preset behaviors.