US 12,437,217 B2
System and method for non-destructive rapid food profiling using artificial intelligence
Woon Siong Alan Lai, Singapore (SG)
Assigned to PROFILEPRINT PTE. LTD., Singapore (SG)
Appl. No. 17/781,786
Filed by PROFILEPRINT PTE. LTD., Singapore (SG)
PCT Filed Dec. 1, 2020, PCT No. PCT/SG2020/050708
§ 371(c)(1), (2) Date Jun. 2, 2022,
PCT Pub. No. WO2021/112762, PCT Pub. Date Jun. 10, 2021.
Claims priority of application No. 10201911636P (SG), filed on Dec. 4, 2019.
Prior Publication US 2023/0029413 A1, Jan. 26, 2023
Int. Cl. G06N 3/00 (2023.01); G06N 5/022 (2023.01); G06N 5/04 (2023.01); G06N 7/00 (2023.01)
CPC G06N 5/04 (2013.01) [G06N 5/022 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A portable apparatus configured to perform non-destructive taste profiling of a food, the portable apparatus comprising:
a receptacle configured to move a sample of the food in a volumetric sampling space, in which the sample includes the food in a non-homogenized form;
a source configured to direct light towards the volumetric sampling space;
an optical device having an input port and an output port, the input port being configured to sense reflectance from at least a part of the sample in the volumetric sampling space, the reflectance characterized by visible-to-near infra-red light, the optical device being configured to output a component of the reflectance through the output port; and
a detector coupled to the output port, the detector being configured to convert the component of the reflectance into captured data, the captured data being characterized by an overtone spectrum of a measure of the reflectance, the overtone spectrum characterized by gradual changes in intensity over a range of wavelengths; and
a computing device coupled to the detector, the computing device being configured to:
execute at least one first machine learning model using the captured data as input, the at least first machine learning model being configured to:
predict at least one facet corresponding to at least one selected wavelength from the overtone spectrum; and
predict a signature data based on a plurality of the at least one facet, wherein the signature data is characteristic of a taste of the food, and
execute at least one second machine learning model using the signature data as input, the at least one second machine learning model being configured to:
predict at least one descriptor; and
predict a signature characteristic of the food using the at least one descriptor, wherein the signature comprises a cultivar or an origin of the food.