US 12,333,802 B2
System and method for three dimensional object counting utilizing point cloud analysis in artificial neural networks
Brad Curtis, Highland, UT (US); Kyle William Armstrong, South Jordan, UT (US); and Porter Reece Jenkins, American Fork, UT (US)
Assigned to DELICIOUS AI LLC, Lehi, UT (US)
Filed by DELICIOUS AI LLC, Lehi, UT (US)
Filed on Oct. 14, 2021, as Appl. No. 17/501,810.
Claims priority of provisional application 63/092,238, filed on Oct. 15, 2020.
Prior Publication US 2022/0121852 A1, Apr. 21, 2022
Int. Cl. G06V 20/00 (2022.01); G06F 18/24 (2023.01); G06F 18/25 (2023.01); G06F 18/40 (2023.01); G06N 3/045 (2023.01); G06Q 10/087 (2023.01); G06V 10/22 (2022.01); G06V 10/40 (2022.01); G06V 10/94 (2022.01); G06V 20/20 (2022.01); G06V 20/64 (2022.01); G01S 7/48 (2006.01); G01S 17/42 (2006.01)
CPC G06V 20/20 (2022.01) [G06F 18/24 (2023.01); G06F 18/251 (2023.01); G06F 18/40 (2023.01); G06N 3/045 (2023.01); G06Q 10/087 (2013.01); G06V 10/225 (2022.01); G06V 10/40 (2022.01); G06V 10/95 (2022.01); G06V 20/64 (2022.01); G01S 7/4802 (2013.01); G01S 17/42 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A system for three-dimensional object counting in a three-dimensional space, the system comprising:
an image capture device configured to obtain at least one 2D image of the three-dimensional space;
a scanner configured to obtain a 3D point cloud of the three-dimensional space;
a processor configured to cooperate with at least one artificial neural network, the at least one artificial neural network being configured to classify and count objects in the three-dimensional space using a monocular architecture of a combination of a 2D image recognition analysis based on the at least one 2D image obtained by the image capture device and a 3D point cloud analysis based on the 3D point cloud obtained by the scanner,
wherein the at least one artificial neural network comprises a distinct first, second, and third artificial neural networks,
wherein the first artificial neural network is configured to receive the at least one 2D image and to output a set of bounding boxes defining an area of each product of an identified product in the three-dimensional space and to define a centroid of at least one identified product,
wherein the second artificial neural network is configured to receive the three-dimensional point cloud and to output a 3D feature vector of the three-dimensional space, and
wherein the third artificial neural network is configured to receive the output of the first and second artificial neural networks and to output a predictions vector comprising a product identification and product count.