Search Result Feedback Apparatus Object Arrangement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional visual search technologies provide poorly targeted and low-accuracy search results due to the lack of effective methods for determining the arrangement order of objects in an image, leading to inefficient feedback to users.
Innovation Solution
A method that determines the arrangement order of objects in an image based on scene intent weight, confidence score, and object relationship score, using image detection processing to prioritize and filter search results, thereby enhancing the accuracy and relevance of the feedback provided to users.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If search results are fed back according to the arrangement order of objects in the image, then the feedback process is simple, but the accuracy of the search result is low
Solution Approach 1:
The patent changes the parameter of object arrangement by introducing multiple scoring dimensions (scene intent weight, confidence score, object relationship score) to reorder objects before feedback. This transforms the simple spatial arrangement into a multi-criteria ranked arrangement, improving search result accuracy without significantly increasing system complexity
Solution Approach 2:
The patent combines multiple evaluation factors (scene intent weight, confidence score, object relationship score) into a composite scoring system for object arrangement. This composite approach integrates different aspects of object importance to achieve more accurate search result feedback while maintaining a unified feedback process
2Productivity
If all objects in the image are searched, then completeness of search is high, but the feedback is poorly targeted and efficiency is low
Solution Approach 1:
The patent applies local quality by differentiating the treatment of different objects based on their individual characteristics (scene intent weight, confidence score, object relationship score). Objects with higher scores receive prioritized feedback, creating localized optimization in the feedback process that improves both efficiency and targeting accuracy
Solution Approach 2:
The patent performs preliminary action by calculating scene intent weight, confidence scores, and object relationship scores for all objects before the feedback process. This pre-ranking of objects based on multiple criteria enables efficient targeted feedback without requiring all objects to be searched equally, improving both productivity and information quality
Data Source
AI summary
In one example method, a first image including M objects is obtained. For N objects in the M objects, when N is greater than or equal to 2, arrangement orders of the N objects is determined, where an arrangement order of any one of the N objects is determined based on at least one of a scene intent weight, a confidence score, or an object relationship score. The scene intent weight is used to indicate a probability that the any object is searched in a scene corresponding to the first image, the confidence score is a similarity between the any object and an image in an image library, and the object relationship score is used to indicate importance of the any object in the first image. Search results of some or all of the N objects are fed back according to the arrangement orders of the N objects.