US 12,283,380 B2
Somatotype identification method, acquisition method of health assessment, apparatus and device
Sifan Wang, Beijing (CN)
Assigned to BOE TECHNOLOGY GROUP CO., LTD., Beijing (CN)
Appl. No. 17/636,356
Filed by BOE TECHNOLOGY GROUP CO., LTD., Beijing (CN)
PCT Filed Apr. 30, 2021, PCT No. PCT/CN2021/091624
§ 371(c)(1), (2) Date Feb. 17, 2022,
PCT Pub. No. WO2022/227042, PCT Pub. Date Nov. 3, 2022.
Prior Publication US 2023/0352178 A1, Nov. 2, 2023
Int. Cl. G16H 50/30 (2018.01); A61B 5/107 (2006.01); G06V 40/10 (2022.01); G16H 50/20 (2018.01)
CPC G16H 50/30 (2018.01) [A61B 5/107 (2013.01); G06V 40/103 (2022.01); G16H 50/20 (2018.01)] 16 Claims
OG exemplary drawing
 
1. A somatotype identification method, wherein the method comprises:
acquiring a body image of a user;
performing a feature extraction on the body image, to obtain a predicted somatotype feature of the user; and
screening a target somatotype category of the user from various somatotype categories according to the predicted somatotype feature;
wherein performing the feature extraction on the body image, to obtain a predicted somatotype feature of the user, comprises:
inputting the body image into a somatotype feature extraction model, to obtain the predicted somatotype feature of the user, wherein the somatotype feature extraction model is obtained by the following steps:
acquiring a sample body image labeled with the somatotype category;
performing a perspective transformation on the sample body image, to obtain a body image set; and
training a somatotype feature extraction model to be trained according to the body image set, to obtain the somatotype feature extraction model for extracting the predicted somatotype feature;
wherein performing a perspective transformation on the sample body image, to obtain a body image set, comprises:
based on vertexes of a target frame region in the sample body image. performing the perspective transformation according to a target zoom range, and combining a plurality of obtained body images into the body image set, wherein the target frame region is an image region where a body of the user is located.