Image Search Query Vectors for Accurate Multi-Component Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image search techniques face challenges in determining image content using computerized methods, making it difficult to effectively match images in a database with a query.
Innovation Solution
Utilizing embeddings to represent image content as vectors and training an image classification engine with a subset of images to improve classification accuracy, employing embeddings to identify similar images and reduce manual labor costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual tagging is used to classify images, then classification accuracy can be achieved, but labor costs and time consumption increase significantly
Solution Approach 1:
The system performs preliminary actions by pre-processing images to extract features and generate embeddings before the actual search operation. This includes resizing, normalization, and feature extraction using pre-trained models, which prepares the data in advance for faster and more accurate classification during search operations.
Solution Approach 2:
The patent introduces embeddings as an intermediary representation between raw images and classification results. Instead of directly comparing images or using manual tagging, the system converts images into vector embeddings that capture semantic meaning, enabling automated classification and search without requiring manual annotation while maintaining high accuracy.
2Measurement precision
If traditional image search methods are used, then simplicity is maintained, but the ability to determine image content accurately deteriorates
Solution Approach 1:
The patent replaces traditional mechanical image comparison methods with machine learning-based embedding systems. Instead of using pixel-by-pixel comparison or simple feature matching, the system uses pre-trained neural network models to generate semantic embeddings, substituting complex automated processing for simpler traditional methods while dramatically improving content determination accuracy.
3Reliability
If more images are manually tagged to improve classification, then classification performance improves, but costs and processing time increase
Solution Approach 1:
The system enables self-service classification by using transfer learning where a pre-trained model on large datasets automatically adapts to specific classification tasks. The model serves itself by leveraging knowledge from pre-training to classify images without requiring extensive manual annotation, achieving high reliability while maintaining processing efficiency through automated feature extraction and comparison.
Data Source
AI summary
A server receives, from a client device, an image search query comprising a plurality of search components. The server generates a plurality of component embedding vectors based on at least a subset of the plurality of search components. The server generates a plurality of query vectors that represent different combinations of two or more of the plurality of component embedding vectors. The server identifies, by accessing a vector database storing image embedding vectors for images, one or more images based on a comparison of the image embedding vectors in the vector database with at least one of the plurality of query vectors. The server transmits, to the client device, information to cause a display, at the client device, of the one or more images in an order determined based on the comparison.