US 12,033,299 B2
Interaction training system for autistic patient using image warping, method for training image warping model, and computer readable storage medium including executions causing processor to perform same
Soonbum Hong, Seoul (KR); and Hyoun-Joong Kong, Seoul (KR)
Assigned to SEOUL NATIONAL UNIVERSITY HOSPITAL, Seoul (KR)
Filed by Seoul National University Hospital, Seoul (KR)
Filed on Sep. 23, 2022, as Appl. No. 17/951,944.
Claims priority of application No. 10-2021-0097811 (KR), filed on Jul. 26, 2021.
Prior Publication US 2023/0123330 A1, Apr. 20, 2023
Int. Cl. G06T 3/00 (2024.01); G06F 3/01 (2006.01); G06T 3/18 (2024.01); G06V 10/40 (2022.01); G06V 10/74 (2022.01); G06V 10/774 (2022.01); G16H 30/40 (2018.01); G16H 50/30 (2018.01); A61B 5/16 (2006.01)
CPC G06T 3/18 (2024.01) [G06F 3/011 (2013.01); G06V 10/40 (2022.01); G06V 10/761 (2022.01); G06V 10/774 (2022.01); G16H 30/40 (2018.01); G16H 50/30 (2018.01); A61B 5/165 (2013.01)] 12 Claims
OG exemplary drawing
 
1. A user interaction training system, comprising:
a photographing device configure to photograph an original image including a first object at a plurality of photographing angles;
an image warping device configured to receive an image including a second object corresponding to a user and the original image including the first object photographed by the photographing device, and to generate a converted image converting the first object into a third object different from the first object based on characteristic information of the second object by using a pre-trained image warping model;
a display device configured to display the generated converted image; and
a sensor configured to measure a stress index of the user,
wherein the image warping model is trained to receive a training image including an arbitrary object, to reconstruct and output the training image based on a feature value extracted from the second object and the stress index,
wherein the sensor is configured to measure the stress index of the user with respect to the converted image generated while the user wears the display device, and
wherein the image warping device reconstructs the converted image by further reflecting the stress index in the image warping model, and if the stress index is less than or equal to a preset first reference stress value, the image warping device reconstructs the converted image such that a first similarity between the third object and the first object becomes greater than a second similarity between the third object and the second object.