IMAGE SEGMENTATION APPARATUS AND METHOD BASED ON DEEP LEARNING

WIPO Patent Application WO/2022/030943

A1

An image segmentation method based on deep learning in an electronic device according to one embodiment of the present document may comprise the operations in which: a processor of the electronic device obtains an input image; the processor converts the input image into a first image having a first resolution and a second image having a second resolution; the processor performs first image processing on the first image using a first deep learning engine; the processor performs second image processing on the second image using a second deep learning engine that is distinguished from the first deep learning engine; and the processor generates an output image on the basis of first region segmentation information according to the first image processing and second region segmentation information according to the second image processing. According to various embodiments, it is possible to provide an electronic device and method capable of solving a quantization error of image segmentation and performance degradation of detailed parts while also maintaining a processing time applicable to the electronic device.

CHUN HYUNGJU (KR)
KIM YOUNGJO (KR)
PARK HYUNHEE (KR)
LEE ARANG (KR)
LIM SUNGJUN (KR)
CHOI JONGBUM (KR)
HAN CHANGSU (KR)

PCT/KR2021/010134

February 10, 2022

August 03, 2021

Click for automatic bibliography generation  

SAMSUNG ELECTRONICS CO LTD (KR)

G06T7/11; G06T7/136; G06T7/174

US20200134772A1	2020-04-30
KR20200068083A	2020-06-15
KR20200063349A	2020-06-05
KR20200017607A	2020-02-19
US10614574B2	2020-04-07

KWON, Hyuk-Rok et al. (KR)

View/Download PDF      PDF Help

Previous Patent: CONDUCTIVE FOAM TAPE FOR ATTACHING DISPLAY PANEL

Next Patent: METHOD FOR DIAGNOSING AUTOIMMUNE ENCEPHALITIS