US 12,106,399 B2
Generative system for the creation of digital images for printing on design surfaces
Rita Cucchiara, Fiorano Modenese (IT); Simone Calderara, Fiorano Modenese (IT); Fabio Lanzi, Fiorano Modenese (IT); Andrea Mariani, Fiorano Modenese (IT); and Elena Pellesi, Fiorano Modenese (IT)
Assigned to DIGITAL DESIGN S.R.L., Fiorano Modenese (IT)
Appl. No. 17/792,737
Filed by DIGITAL DESIGN S.R.L., Fiorano Modenese (IT)
PCT Filed Jan. 14, 2021, PCT No. PCT/IB2021/050248
§ 371(c)(1), (2) Date Jul. 14, 2022,
PCT Pub. No. WO2021/144728, PCT Pub. Date Jul. 22, 2021.
Claims priority of application No. 102020000000664 (IT), filed on Jan. 15, 2020.
Prior Publication US 2023/0045937 A1, Feb. 16, 2023
Int. Cl. G06T 11/00 (2006.01); G06N 3/045 (2023.01)
CPC G06T 11/001 (2013.01) [G06N 3/045 (2023.01)] 18 Claims
OG exemplary drawing
 
1. A generative system for the creation of digital images for printing on design surfaces, the generative system comprising:
at least one training dataset including a plurality of sample images for said printing on design surfaces; and
at least one generative adversarial network including a generator and a discriminator, wherein said generator receives noise at input and is trained to generate at output starting from said noise at least a new artificially generated image adapted to be used for printing on design surfaces; and
said discriminator receives at input said at least one new artificially generated image and is trained to compare and distinguish said new image generated by said sample images of the training dataset,
at least one preliminary progressive-growth training phase of said generator and of said discriminator including at least the following steps:
a) using a first convolutive layer of said generator for the generation of at least one new artificially generated image at a first predefined resolution;
b) using a first convolutive layer of said discriminator to compare and distinguish said new image generated by said sample images of the training dataset:
c) adding by regular steps an additional convolutive layer to said generator and to said discriminator;
d) using the convolutive layers of said generator for the generation of at least one new artificially generated image at a predefined resolution higher than the resolution of the previously generated images:
e) using the convolutive layers of said discriminator to compare and distinguish said new image generated by said sample images of the training dataset: and
repeating the steps c) to e) until a predefined number of convolutive layers are reached, for the generation of at least one artificially generated image at a final predefined resolution.