Untagged Image Search via Text-to-Visual Query Bridge
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional image search techniques are restrictive and limited, as text-based searches rely on keyword tags and visually-based searches require high-quality or relevant query images, making it difficult to find images without proper tagging or query images.
Innovation Solution
A computing device conducts a text-based search on untagged or partially tagged databases, generating a subset of example images to create a visually-based query, allowing for the retrieval of visually similar images from both tagged and untagged databases using publicly accessible image search engines.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If text-based image search is used on untagged image databases, then the search scope is expanded, but the search accuracy deteriorates because traditional text-based searches rely on keyword tags
Solution Approach 1:
The patent introduces an intermediary process that uses a publicly accessible image search engine to generate example images from the text query, which then serves as a bridge to perform visually-based search on the untagged database. This intermediary step enables the system to overcome the lack of keyword tags while maintaining search accuracy through visual similarity comparison.
Solution Approach 2:
The patent replaces the traditional mechanical keyword-matching system with a visually-based search mechanism. Instead of relying on text-based keyword tags, the system uses image processing and visual similarity algorithms to search the untagged database, substituting the mechanical tag-matching approach with a more sophisticated visual analysis system.
2Measurement precision
If visually-based image search is used, then search accuracy is improved, but the requirement for high-quality query images increases operational complexity
Solution Approach 1:
The system performs self-service by automatically generating example images from the text query using a publicly accessible image search engine. This eliminates the need for users to manually provide query images, and the system autonomously completes the image generation and visual search process, reducing operational complexity while maintaining high search accuracy.
Solution Approach 2:
The system performs preliminary action by pre-generating example images from the text query before conducting the visual search on the untagged database. This preliminary step of obtaining reference images automatically allows the system to execute the visually-based search without requiring users to prepare query images, thereby simplifying the user operation while preserving search accuracy.
3Ease of operation
If traditional text-based search is used, then ease of operation is maintained, but the search results are limited by database tagging quality
Solution Approach 1:
The patent merges text-based search and visually-based search into a unified system. It combines the ease of text-based querying with the power of visually-based search by using the text query to generate example images, which then drive the visual search process. This integration allows users to input simple text queries while achieving comprehensive search results that overcome database tagging limitations.
Solution Approach 2:
The system achieves universality by enabling a single text-based interface to perform both text-based and visually-based search functions. The same text query input can trigger automatic example image generation and subsequent visual search, allowing the system to handle both tagged and untagged databases uniformly without requiring different user interfaces or operations.
Data Source
AI summary
In various implementations, a personal asset management application is configured to perform operations that facilitate the ability to search multiple images, irrespective of the images having characterizing tags associated therewith or without, based on a simple text-based query. A first search is conducted by processing a text-based query to produce a first set of result images used to further generate a visually-based query based on the first set of result images. A second search is conducted employing the visually-based query that was based on the first set of result images received in accordance with the first search conducted and based on the text-based query. The second search can generate a second set of result images, each having visual similarity to at least one of the images generated for the first set of result images.