3D Model Search Using Image Representation Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for searching three-dimensional models using text-based terms or words are inefficient due to language differences and varying terminology, leading to inadequate results, as users and content creators may use different terms or languages to describe models.
Innovation Solution
A hybrid machine-learning model is employed to generate a representation vector from a two-dimensional image, using supervised and unsupervised learning layers, which projects features into multiple dimensions, enabling efficient comparison and search for similar three-dimensional models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If text-based search terms are used to search for three-dimensional models, then the search process is simple, but the search accuracy deteriorates due to language differences and varying terminology
Solution Approach 1:
The patent introduces an image representation vector as an intermediary between the user's search query and the three-dimensional model database. Instead of directly matching text terms with model metadata, the system converts the search image into a vector representation and compares it with vectors of model snapshots, thereby mediating the search process to overcome language and terminology barriers
Solution Approach 2:
The patent replaces the mechanical text-based search system with an image-based machine learning system. The hybrid machine learning model processes images and generates representation vectors, substituting the traditional text-matching mechanism with a visual recognition and vector comparison approach that is insensitive to language differences
2Measurement precision
If a hybrid machine-learning model with multiple layers is used to generate representation vectors, then search accuracy improves, but computational complexity increases
Solution Approach 1:
The patent segments the machine learning model into two distinct components: a hybrid machine learning model for generating representation vectors and a separate vector comparison module for searching. This segmentation allows the complex model to be used only during the vector generation phase, while the search phase uses simpler vector operations, thereby distributing computational complexity across different stages
Solution Approach 2:
The patent performs preliminary action by pre-computing and storing representation vectors for three-dimensional model snapshots in the database. This allows the complex hybrid machine learning model to be executed only once during data preparation, rather than during each search operation. Subsequent searches only require comparing pre-generated vectors, significantly reducing real-time computational complexity
Data Source
AI summary
One or more three-dimensional models that corresponds to at least one two-dimensional image are determined by receiving image data that corresponds to the at least one two-dimensional image, generating features based on the image data corresponding to the at least one two-dimensional image, generating a representation vector for the at least one two-dimensional image by transforming the features into a predetermined amount of numerical representations corresponding to the features, and outputting the representation vector for the at least one two-dimensional image to facilitate a search query for the one or more three-dimensional models associated with the at least one two-dimensional image.