US 12,217,400 B2
Method for image processing by means of a neural network and method for training a neural network
Annika Baumann, Hamburg (DE)
Assigned to Basler AG, Ahrensburg (DE)
Appl. No. 17/760,707
Filed by Basler AG, Ahrensburg (DE)
PCT Filed Sep. 17, 2020, PCT No. PCT/EP2020/075964
§ 371(c)(1), (2) Date Mar. 15, 2022,
PCT Pub. No. WO2021/053066, PCT Pub. Date Mar. 25, 2021.
Claims priority of application No. 10 2019 125 000.5 (DE), filed on Sep. 17, 2019.
Prior Publication US 2022/0392074 A1, Dec. 8, 2022
Int. Cl. G06T 5/70 (2024.01); G06T 7/10 (2017.01); G06V 10/30 (2022.01); G06V 10/44 (2022.01); G06V 10/82 (2022.01)
CPC G06T 5/70 (2024.01) [G06T 7/10 (2017.01); G06V 10/30 (2022.01); G06V 10/454 (2022.01); G06V 10/82 (2022.01); G06T 2207/20021 (2013.01); G06T 2207/20084 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method for processing an image using a neural network, comprising:
dividing the image into a plurality of image regions;
grouping at least two image regions of the plurality of image regions into at least two groups based on one or more properties of the plurality of image regions, wherein each of the plurality of image regions is assigned to exactly one group or to no group and image regions with the same or similar properties are assigned to the same group;
computing partial results for the image regions of the at least two groups by:
applying at least two kernels in a layer of the neural network to the image regions of the at least two groups, wherein at most one kernel is applied to the image regions of each group; and
generating a result matrix of a layer of the neural network based on the partial results for the image regions of the at least two groups, wherein a data volume of the result matrix is less than a data volume that would be produced by applying each of the at least two kernels to each image region of the plurality of image regions.