基于自然语言查询的个性化图像和视频搜索的方法及其装置
By anonymizing natural language queries and applying multi-model scoring, combined with facial information, the problem of the inability of existing technologies to effectively handle proper name queries is solved, achieving accurate personalized image and video search.
CN116868181BActive Publication Date: 2026-07-17SAMSUNG ELECTRONICS CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SAMSUNG ELECTRONICS CO LTD
- Filing Date
- 2022-01-07
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing image search technologies cannot effectively handle queries based on natural language, especially queries for proper names, and malfunctions in different detectors lead to inaccurate results.
Method used
By anonymizing proper nouns in natural language queries by replacing them with general terms, and by using trained scoring and re-scoring models in conjunction with facial information, initial and incremental scores are calculated to ultimately determine the probability of faces in an image.
Benefits of technology
It enables personalized image search based on natural language queries, improving the accuracy and applicability of the results, and effectively retrieving images and videos containing proper names.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN116868181B_ABST
Abstract
一种个性化图像检索的方法包括：获得包括名称的自然语言查询；用通用术语替换自然语言查询中的名称，以提供匿名化的查询和带名称的实体信息；使用输入匿名化的查询和多个图像的经训练的评分模型，来获得对应于多个图像的多个初始排名得分和多个关注权重；使用输入多个关注权重和带名称的实体信息的重新评分模型，来获得对应于多个图像的多个增量得分；以及通过基于多个增量得分修改多个初始排名得分，来获得多个最终排名得分。经训练的评分模型执行基于语义的搜索，并且重新评分模型确定在多个图像中检测到的面部对应于名称的概率。
Need to check novelty before this filing date? Find Prior Art