US 12,326,916 B2
Method of generating image recognition model and electronic device using the method
Ching-Wen Cheng, Hsin-Chu (TW); Yen-Chun Huang, Hsin-Chu (TW); Yi-Fan Liou, Hsin-Chu (TW); and Kui-Ting Chen, Hsin-Chu (TW)
Assigned to Coretronic Corporation, Hsin-Chu (TW)
Filed by Coretronic Corporation, Hsin-Chu (TW)
Filed on Oct. 29, 2021, as Appl. No. 17/513,896.
Claims priority of application No. 202011402819.2 (CN), filed on Dec. 2, 2020.
Prior Publication US 2022/0171994 A1, Jun. 2, 2022
Int. Cl. G06F 18/214 (2023.01); G06N 20/00 (2019.01); G06V 10/22 (2022.01); G06V 10/26 (2022.01); G06V 10/40 (2022.01); G06V 40/16 (2022.01)
CPC G06F 18/214 (2023.01) [G06V 10/22 (2022.01); G06V 10/267 (2022.01); G06V 10/40 (2022.01); G06N 20/00 (2019.01); G06V 40/16 (2022.01)] 10 Claims
OG exemplary drawing
 
1. A method of generating an image recognition model, comprising:
obtaining a source image;
cutting a first image out of a first region of the source image to generate a cut source image, wherein the first image is different from the cut source image, and the first image and the cut source image form the source image;
pre-training a preliminary image recognition model according to first feature data and first label data to generate a pre-trained preliminary image recognition model, wherein the first feature data is associated with the cut source image, and the first label data is generated by encoding the first image; and
fine-tuning the pre-trained preliminary image recognition model to generate the image recognition model.