-
公开(公告)号：US20220058448A1
公开(公告)日：2022-02-24
申请号：US17419427
申请日：2019-04-09
摘要： Disclosed herein is a method of determining a user profile based on a set of user-selected images, a method of selecting images from an image database of digital images based on a user profile, a computer system and a computer program product. The method of determining a user profile comprises obtaining a set of reference images, wherein each of the reference images is associated with a category from a plurality of categories; determining a sample feature vector for a sample image and a reference feature vector for each of the reference images, wherein the feature vector of an image is associated to features of the image; determining a similarity metric between the sample image and each of the reference images based on the sample feature vector and the reference feature vectors; selecting nearest reference images for each category, wherein the similarity metric between the sample image and a nearest reference image meets a minimum assignment similarity criterion and a maximum assignment similarity criterion; and determining the user profile by calculating an assignment probability for each category based on the similarity metrics between the sample image and the nearest reference images of the respective category.
-
公开(公告)号：US11914638B2
公开(公告)日：2024-02-27
申请号：US17419427
申请日：2019-04-09
IPC分类号： G06F16/56 , G06F16/55 , G06F16/583 , G06V10/94 , G06N3/02 , G06F18/2413 , G06F18/22 , G06F18/2415 , G06F18/2431 , G06V10/74 , G06V10/762 , G06V10/764 , G06V20/00
CPC分类号： G06F16/56 , G06F16/55 , G06F16/583 , G06F18/22 , G06F18/2415 , G06F18/2431 , G06F18/24147 , G06N3/02 , G06V10/761 , G06V10/763 , G06V10/764 , G06V10/95 , G06V20/35
摘要： Disclosed herein is a method of determining a user profile based on a set of user-selected images, a method of selecting images from an image database of digital images based on a user profile, a computer system and a computer program product. The method of determining a user profile comprises obtaining a set of reference images, wherein each of the reference images is associated with a category from a plurality of categories; determining a sample feature vector for a sample image and a reference feature vector for each of the reference images, wherein the feature vector of an image is associated to features of the image; determining a similarity metric between the sample image and each of the reference images based on the sample feature vector and the reference feature vectors; selecting nearest reference images for each category, wherein the similarity metric between the sample image and a nearest reference image meets a minimum assignment similarity criterion and a maximum assignment similarity criterion; and determining the user profile by calculating an assignment probability for each category based on the similarity metrics between the sample image and the nearest reference images of the respective category.
-
搜索结果
2 条记录 (耗时 0.012 秒)
