US 11,941,838 B2
Methods, apparatuses, devices and storage medium for predicting correlation between objects
Chunya Liu, Singapore (SG); Xuesen Zhang, Singapore (SG); Bairun Wang, Singapore (SG); and Jinghuan Chen, Singapore (SG)
Assigned to SENSETIME INTERNATIONAL PTE. LTD., Singapore (SG)
Filed by SENSETIME INTERNATIONAL PTE. LTD., Singapore (SG)
Filed on Jun. 30, 2021, as Appl. No. 17/363,244.
Application 17/363,244 is a continuation of application No. PCT/IB2021/054778, filed on Jun. 1, 2021.
Claims priority of application No. 10202102717U (SG), filed on Mar. 17, 2021.
Prior Publication US 2022/0301219 A1, Sep. 22, 2022
Int. Cl. G06T 7/73 (2017.01); G06K 9/00 (2022.01); G06V 20/40 (2022.01); G06V 40/10 (2022.01); G06V 40/16 (2022.01)
CPC G06T 7/73 (2017.01) [G06V 20/46 (2022.01); G06V 40/107 (2022.01); G06V 40/161 (2022.01); G06T 2207/30201 (2013.01); G06T 2210/12 (2013.01)] 18 Claims
OG exemplary drawing
 
1. A method for predicting correlation between objects, comprising:
detecting a first object, a second object, and a third object involved in a target image, wherein the first object and the second object represent different body parts, and the third object indicates a body object;
determining a joint bounding box surrounding the first object, the second object, and the third object; and
predicting correlation between the first object and the second object based on a region corresponding to the joint bounding box in the target image;
further comprising:
determining correlated objects involved in the target image based on a prediction result of the correlation between the first object and the second object;
wherein the prediction result of the correlation comprises a correlation prediction score,
determining correlated objects involved in the target image based on the prediction result of the correlation between the first object and the second object comprises:
selecting each from a plurality of object pairs, as a current object pair, in an order of respective correlation prediction scores of the object pairs from high to low, wherein the plurality of object pairs are obtained by combining each detected first object with each detected second object; and
for the current object pair,
based on determined correlated objects involved in the target image, determining a number of second objects that are correlated with the first object in the current object pair as a first number, and determining a number of first objects that are correlated with the second object in the current object pair as a second number; and
in response to that the first number is lower than a first preset threshold, and the second number is lower than a second preset threshold, determining the first object and the second object in the current object pair as correlated objects involved in the target image.