US 12,407,854 B2
Artificial intelligence (AI) encoding apparatus and method and AI decoding apparatus and method for region of object of interest in image
Heechul Yang, Suwon-si (KR); Hyunkwon Chung, Suwon-si (KR); and Inhak Na, Suwon-si (KR)
Assigned to SAMSUNG ELECTRONICS CO., LTD., Suwon-si (KR)
Filed by SAMSUNG ELECTRONICS CO., LTD., Suwon-si (KR)
Filed on May 9, 2023, as Appl. No. 18/195,221.
Application 18/195,221 is a continuation of application No. PCT/KR2021/013363, filed on Sep. 29, 2021.
Claims priority of application No. 10-2020-0148696 (KR), filed on Nov. 9, 2020; and application No. 10-2020-0173500 (KR), filed on Dec. 11, 2020.
Prior Publication US 2023/0276070 A1, Aug. 31, 2023
Int. Cl. H04N 19/59 (2014.01); H04N 19/124 (2014.01); H04N 19/169 (2014.01)
CPC H04N 19/59 (2014.11) [H04N 19/124 (2014.11); H04N 19/188 (2014.11)] 10 Claims
OG exemplary drawing
 
1. An artificial intelligence (AI) encoding apparatus comprising:
a memory storing one or more instructions; and
a processor configured to execute the one or more instructions stored in the memory to:
identify an object region of interest in an original image,
obtain, from the original image, a first original part image comprising the object region of interest, and a second original part image comprising a non-interest region,
obtain a plurality of first images by performing AI scaling on the first original part image and the second original part image through a scaling neural network (NN),
wherein the performing AI scaling comprises applying the first original part image to the scaling NN that is configured with first NN setting information selected from among a plurality of NN setting information to increase a size of the first original part image or to maintain the size of the first original part image, and applying the second original part image to the scaling NN that is configured with second NN setting information selected from among the plurality of NN setting information to decrease a size of the second original part image,
generate image data by encoding the plurality of first images, and
transmit the image data, and AI data comprising information related to the AI scaling.