US 11,869,228 B2
System and a method for generating an image recognition model and classifying an input image
Guangda Li, Singapore (SG); Thanh Tung Todd Cao, Singapore (SG); Zhenhua Wang, Singapore (SG); and Xin Ji, Singapore (SG)
Assigned to VISENZE PTE LTD, Singapore (SG)
Appl. No. 17/763,606
Filed by VISENZE PTE LTD, Singapore (SG)
PCT Filed Nov. 6, 2020, PCT No. PCT/SG2020/050642
§ 371(c)(1), (2) Date Mar. 24, 2022,
PCT Pub. No. WO2022/098295, PCT Pub. Date May 12, 2022.
Prior Publication US 2023/0237769 A1, Jul. 27, 2023
Int. Cl. G06V 10/764 (2022.01); G06V 10/77 (2022.01); G06N 3/082 (2023.01)
CPC G06V 10/764 (2022.01) [G06N 3/082 (2013.01); G06V 10/7715 (2022.01)] 14 Claims
OG exemplary drawing
 
1. A method of generating an image recognition model for recognizing an input image, comprising:
appending at least one feature extraction layer to the image recognition model;
extracting a plurality of feature vectors from a set of predetermined images;
grouping the plurality of feature vectors into a plurality of categories;
clustering the plurality of feature vectors of each of the plurality of categories into at least one cluster;
determining at least one centroid for each of the at least one cluster, wherein each of the at least one cluster comprises at least one centroid, wherein each of the at least one centroid is represented by a feature vector;
generating a classification layer based on the feature vector of each of the at least one centroid of the plurality of categories;
appending the classification layer to the image recognition model; and
receiving a plurality of selected images in one of the plurality of categories, extracting a plurality of new feature vectors of the plurality of selected images, adding the plurality of new feature vectors to the plurality of feature vectors of the one of the plurality of categories to form a new set of feature vectors, re-clustering the new set of feature vectors into at least one cluster, re-generating at least one centroid for each of the at least one cluster, wherein each of the at least one centroid is represented by a feature vector, and appending the feature vector of each of the at least one centroid for each of the at least one cluster to the classification layer of the image recognition model.