US 12,333,855 B1
Automated system for generating personalized eyewear recommendations based on fit and style within a purchasing system
Mehdi Faraji, Vancouver (CA); Gregory Matthews, Calgary (CA); Niharika Khandelwal, Edmonton (CA); Gaurav Mishra, Edmonton (CA); Maryam Beikmohammadloo, Edmonton (CA); Amir Narimani, Edmonton (CA); and Ava Sehat-Niaki, Edmonton (CA)
Assigned to Eyevious Style Incorporated, Calgary (CA)
Filed by Eyevious Style Incorporated, Calgary (CA)
Filed on Jun. 11, 2021, as Appl. No. 17/345,356.
Claims priority of provisional application 63/037,835, filed on Jun. 11, 2020.
Int. Cl. G06T 7/70 (2017.01); G06F 18/2113 (2023.01); G06Q 30/0251 (2023.01); G06V 40/16 (2022.01)
CPC G06V 40/171 (2022.01) [G06F 18/2113 (2023.01); G06Q 30/0269 (2013.01); G06T 7/70 (2017.01); G06T 2207/30201 (2013.01)] 60 Claims
OG exemplary drawing
 
1. A system for image analysis, the system comprising a processor, memory accessible by the processor, and computer program instructions stored in the memory and executable by the processor to perform:
obtaining an image of a face or head of a person;
generating face features from the image of the face or head of the person;
retrieving a plurality of object features of a plurality of objects from a database of object features of the plurality of objects, wherein the plurality of objects comprises products purchasable by a user, and wherein the database of object features comprises parameters computed based on physical properties of the plurality of objects;
generating a plurality of combined face and object features from the face features and the plurality of object features;
generating a ranking of combined face and object features, wherein generating the ranking comprises generating a ranking of object fit relative to the face features based on the plurality of combined face and object features;
receiving a preference classification from the person, wherein the preference classification comprises real-time iterative feedback relating to the plurality of object features, wherein the preference classification is determined through an adaptive learning process that uses a heuristic that takes into account preference classification information received from the person on all prior object feature information presented to the person relating to the plurality of object features, and wherein the heuristic is based on a classification confidence of a previous training state of the adaptive learning process for one or more individual object features of the plurality of objects;
selecting at least one object based on the ranking of combined face and object features and the preference classification;
generating an interactive display of the selected at least one object to the person; and
providing user interaction of the interactive display to provide purchase of the selected object.