Capsule Neural Image Retrieval Using Distilled Image Sets
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image retrieval systems struggle to provide similar images quickly, efficiently, and accurately, often requiring complex models that increase operational costs.
Innovation Solution
A method utilizing a capsule neural network model to distill an original image set into a distilled image set, extracting features, determining similarities, and acquiring matching images, which reduces the need for complex models and lowers operational costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex models are used for image retrieval, then retrieval accuracy is improved, but operational costs increase
Solution Approach 1:
The patent creates a distilled image set that serves as a simplified copy of the original image set. This distilled set contains only the most representative images, allowing the system to perform retrieval operations on a smaller subset while maintaining high accuracy. The copying principle resolves the contradiction by enabling accurate retrieval without processing the entire original dataset, thus reducing operational costs.
Solution Approach 2:
The patent extracts essential features and representative images from the original image set to create a distilled version. By taking out only the most important elements (key images and their features), the system achieves accurate retrieval with reduced computational resources, thereby lowering operational costs while maintaining retrieval precision.
2Measurement precision
If complex models are used for image retrieval, then retrieval accuracy is improved, but system complexity increases
Solution Approach 1:
Instead of using complex models on the entire original image set, the patent creates a simplified copy (distilled image set) that preserves the essential retrieval characteristics. This approach maintains high retrieval accuracy while significantly reducing model and system complexity by working with a condensed representation of the data.
Solution Approach 2:
The patent extracts only the necessary components (representative images and key features) from the original set, discarding redundant information. This extraction process simplifies the system by removing unnecessary complexity while retaining the essential elements needed for accurate retrieval.
3Reliability
If the entire original image set is processed, then retrieval completeness is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary processing to create a distilled image set before the actual retrieval operation. This pre-processing step identifies and extracts the most relevant images in advance, so that during retrieval, the system only needs to search through the pre-selected subset. This preliminary action ensures retrieval completeness for important cases while dramatically reducing processing time.
Solution Approach 2:
The distilled image set serves as a pre-prepared copy of the essential content from the original set. By having this condensed version ready in advance, the system can perform rapid retrieval operations without processing the entire original dataset, thus maintaining completeness for key results while reducing time loss.
Data Source
AI summary
Embodiments of the present disclosure relate to a method, an electronic device, and a computer program product for acquiring an image. The method includes distilling an original image set through a capsule neural network model to generate a distilled image set, wherein the distilled image set includes a plurality of distilled images. The method further includes acquiring a first feature of a first image through the capsule neural network model. The method further includes acquiring a plurality of distilling features of the plurality of distilled images respectively through the capsule neural network model. The method further includes determining a plurality of similarities between the first feature and the plurality of distilling features respectively. The method further includes acquiring at least one original image matching the first image based on the plurality of similarities.