US 11,657,520 B2
Electronic device and method for controlling same
Daehyun Ban, Suwon-si (KR); Woojin Park, Suwon-si (KR); and Seongwon Han, Suwon-si (KR)
Assigned to Samsung Electronics Co., Ltd., Suwon-si (KR)
Appl. No. 16/771,922
Filed by Samsung Electronics Co., Ltd., Suwon-si (KR)
PCT Filed Jan. 3, 2019, PCT No. PCT/KR2019/000101
§ 371(c)(1), (2) Date Jun. 11, 2020,
PCT Pub. No. WO2019/168264, PCT Pub. Date Sep. 6, 2019.
Claims priority of application No. 10-2018-0023986 (KR), filed on Feb. 27, 2018.
Prior Publication US 2020/0402251 A1, Dec. 24, 2020
Int. Cl. G06T 7/50 (2017.01); G06T 7/593 (2017.01); G06N 20/00 (2019.01); G06V 20/64 (2022.01); G06F 18/21 (2023.01); G06F 18/24 (2023.01); G06V 10/70 (2022.01); G06V 10/82 (2022.01)
CPC G06T 7/50 (2017.01) [G06F 18/21 (2023.01); G06F 18/24 (2023.01); G06N 20/00 (2019.01); G06T 7/593 (2017.01); G06V 10/70 (2022.01); G06V 10/82 (2022.01); G06V 20/64 (2022.01); G06T 2207/10012 (2013.01); G06T 2207/10028 (2013.01); G06T 2207/20081 (2013.01)] 10 Claims
OG exemplary drawing
 
1. An electronic device comprising a learning model trained according to an artificial intelligence algorithm, the electronic device comprising:
an input unit; and
a processor configured to:
based on a two-dimensional image including at least one object being received via the input unit, obtain first depth information regarding the at least one object and information regarding a type of the at least one object by applying the two-dimensional image to a first learning model, the first depth information comprising depth data according to the type of the at least one object,
obtain second depth information regarding the at least one object by applying the first depth information and ground truth depth data of the at least one object to a second learning model, and
obtain three-dimensional information regarding the two-dimensional image based on the second depth information, the three-dimensional information comprising a distance from the at least one object, a relative position and an XYZ coordinate value based on the second depth information,
wherein the first depth information comprises a depth value corresponding to each of a plurality of pixels included in the at least one object,
wherein the processor is further configured to obtain the second depth information by adjusting the depth value according to a Euclidean distance between the depth value and the ground truth depth data corresponding to each of the plurality of pixels by applying the depth value to the second learning model,
wherein the depth data included in the first learning model is a representative depth value generalized according to the type of the at least one object,
wherein the ground truth depth data included in the second learning model is a ground truth depth value obtained by capturing the at least one object with a stereo camera, and
wherein the processor is further configured to obtain the second depth information by adjusting the depth value of the first depth information to minimize an average of a plurality of Euclidean distances obtained by using the second learning model, the second depth information including information regarding the type of the at least one object based on the depth value included in the second depth information which is more sophisticated than the depth value included in the first depth information.