Image Retrieval Reliability Verification via Dual Learning Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image retrieving devices lack the ability to verify the reliability of retrieved images, making it uncertain whether the subject in the retrieved gallery images matches the query image with high probability.
Innovation Solution
The device employs two learning models: a first model to acquire feature vectors for query and gallery images, and a second model to assess the reliability of retrieval by determining the likelihood of gallery images containing the query image's subject, allowing for the specification of retrieval reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single learning model is used for image retrieval, then the device complexity is low, but the reliability of retrieval cannot be verified
Solution Approach 1:
The patent divides the retrieval system into two separate learning models: a first learning model for feature extraction and a second learning model for reliability assessment. This segmentation allows each model to specialize in its function, enabling reliability verification while maintaining manageable complexity through clear functional separation.
Solution Approach 2:
The second learning model acts as an intermediary that assesses the reliability of retrieval results generated by the first learning model. This intermediary component provides independent verification without fundamentally altering the core retrieval mechanism, thus improving reliability while adding only moderate complexity.
2Reliability
If multiple learning models are used to verify retrieval reliability, then the reliability of retrieval can be checked, but the device complexity increases
Solution Approach 1:
Both learning models share common functionality in processing image features and generating retrieval results. The first model handles both feature extraction and initial retrieval, while the second model performs reliability assessment on the same retrieval results. This multi-functionality reduces overall system complexity compared to having completely separate systems.
Solution Approach 2:
The system uses its own retrieval mechanism (first learning model) to generate results that are then assessed by another component (second learning model) within the same system. This self-service approach allows the system to verify its own outputs without requiring external verification systems, thereby controlling complexity.
Data Source
AI summary
An image retrieving device includes processing circuitry configured to give a query image that is an image to be identified to a first learning model, acquire a feature vector of the query image from the first learning model, give each of a plurality of gallery images to the first learning model, and acquire a feature vector of each of the gallery images from the first learning model; give the query image to a second learning model, and acquire, from the second learning model, reliability of retrieval when K gallery images having a relatively high possibility of including a subject included in the query image are retrieved from the plurality of the gallery images; retrieve the K gallery images from the plurality of the gallery images; and specify the reliability of retrieval.