US 12,340,547 B2
Electronic device and method for detecting target objects
Juyong Song, Suwon-si (KR); Byeongcheol Kang, Suwon-si (KR); Youngchul Sohn, Suwon-si (KR); Ilgu Kang, Suwon-si (KR); and Sunghyun Choi, Suwon-si (KR)
Assigned to Samsung Electronics Co., Ltd., Suwon-si (KR)
Filed by Samsung Electronics Co., Ltd., Suwon-si (KR)
Filed on Dec. 15, 2022, as Appl. No. 18/066,695.
Application 18/066,695 is a continuation of application No. PCT/KR2022/019101, filed on Nov. 29, 2022.
Claims priority of application No. 10-2021-0190339 (KR), filed on Dec. 28, 2021; and application No. 10-2022-0009236 (KR), filed on Jan. 21, 2022.
Prior Publication US 2023/0206581 A1, Jun. 29, 2023
Int. Cl. G06V 10/22 (2022.01); G06V 10/774 (2022.01)
CPC G06V 10/22 (2022.01) [G06V 10/774 (2022.01)] 20 Claims
OG exemplary drawing
 
1. An electronic device configured to detect a target object, the electronic device comprising:
a display;
a memory storing one or more instructions; and
a processor configured to execute the one or more instructions stored in the memory to:
apply an image including a first object and a second object to an artificial intelligence model to obtain output data including first output data regarding the first object and second output data regarding the second object,
wherein the first output data regarding the first object includes data regarding a plurality of first candidate boxes corresponding to the first object,
wherein the second output data regarding the second object includes data regarding a plurality of second candidate boxes corresponding to the second object, and
wherein the processor is further configured to:
determine a first detection box corresponding to the first object by selecting one of the plurality of first candidate boxes,
determine a second detection box corresponding to the second object by selecting one of the plurality of second candidate boxes,
wherein the first detection box is used for detecting the first object in the image,
wherein the second detection box is used for detecting the second object in the image, and
wherein the artificial intelligence model is trained to:
obtain a training image including a third object and a fourth object adjacent to the third object,
output a plurality of third candidate boxes corresponding to the third object and a plurality of fourth candidate boxes corresponding to the fourth object, by using the training image, and
determine sizes of the plurality of third candidate boxes and the plurality of fourth candidate boxes by using a loss function for reducing a size difference between the third candidate boxes and the fourth candidates boxes.