US 12,354,323 B2
Image processing method and apparatus, device, medium and program product
Ruoyu Guo, Beijing (CN); Yuning Du, Beijing (CN); Shengyu Wei, Beijing (CN); Shuilong Dong, Beijing (CN); Qiwen Liu, Beijing (CN); Qiao Zhao, Beijing (CN); Ran Bi, Beijing (CN); Xiaoguang Hu, Beijing (CN); Dianhai Yu, Beijing (CN); and Yanjun Ma, Beijing (CN)
Assigned to BEIJING BAIDU NETCOM SCIENCE TECHNOLOGY CO., LTD., Beijing (CN)
Filed by Beijing Baidu Netcom Science Technology Co., Ltd., Beijing (CN)
Filed on Dec. 9, 2022, as Appl. No. 18/078,635.
Claims priority of application No. 202111507276.5 (CN), filed on Dec. 10, 2021.
Prior Publication US 2023/0186599 A1, Jun. 15, 2023
Int. Cl. G06V 10/74 (2022.01); G06V 10/75 (2022.01)
CPC G06V 10/761 (2022.01) [G06V 10/751 (2022.01)] 20 Claims
OG exemplary drawing
 
1. An image processing method, comprising:
acquiring an original image to be placed in an index library, wherein the original image corresponds to label information;
performing image augmentation on the original image to obtain at least one augmented image;
performing subject detection on the original image and the at least one augmented image to obtain an original detection frame in the original image and an augmented detection frame in the at least one augmented image;
determining whether the original detection frame and the augmented detection frame belong to a same subject;
in response to the original detection frame and the augmented detection frame belonging to the same subject, determining a target subject frame in the original image according to the augmented detection frame;
cutting out the target subject frame from the original image, and storing at least one cut target subject frame correspondingly with the label information of the original image;
performing feature extraction on the target subject frame stored correspondingly with the label information to obtain an extracted feature; and
adding the extracted feature and the label information to the index library;
wherein in response to the original detection frame and the augmented detection frame belonging to the same subject, determining the target subject frame in the original image according to the augmented detection frame comprises:
determining a position of the target subject frame according to an original frame position of the original detection frame and an augmented frame position of the augmented detection frame that belong to the same subject; and
determining the target subject frame in the original image according to the position of the target subject frame, wherein the determined target subject frame comprises the original detection frame and the augmented detection frame.