3D Model Component Search with 2D Image Embeddings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Identifying specific components within large collections of three-dimensional models, such as CAD models, is challenging due to the difficulty in locating and searching for particular components within vast data stores when users are unsure of their location or how to describe them effectively.
Innovation Solution
A computing device generates representation vectors from 2-D images of 3-D models and components using a hybrid machine-learning model, indexes these vectors, and compares them to query embeddings to identify and present similar components or models, facilitating efficient searching and retrieval.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If users search for specific components in large collections of 3-D models using traditional methods, then they can locate models in data stores, but the search process becomes difficult and inefficient when users are unsure of component location or cannot describe them effectively
Solution Approach 1:
The patent creates 2-D image representations (copies) of 3-D model components that can be easily processed and compared. These 2-D images serve as simplified copies that retain essential visual characteristics while reducing computational complexity, enabling efficient similarity search without requiring users to know exact component locations or descriptions.
Solution Approach 2:
The patent replaces traditional keyword-based or location-based search mechanisms with a machine learning-based image similarity comparison system. The hybrid machine learning model automatically compares visual features of query components against stored components, substituting manual search operations with automated visual recognition and matching.
2Measurement precision
If the system processes and compares detailed 3-D model data directly, then accurate component identification is achieved, but computational resources and processing time increase significantly
Solution Approach 1:
The patent extracts essential visual features from complete 3-D model data by generating representative 2-D images. This extraction process isolates the most discriminative visual characteristics needed for component identification while discarding redundant information, thereby reducing computational resource requirements while maintaining identification accuracy.
Solution Approach 2:
The patent transforms 3-D model data into 2-D image representations, changing the dimensional parameters and data structure. This parameter transformation reduces the complexity of the data while preserving essential visual features, enabling efficient comparison and matching with lower computational overhead.
3Measurement precision
If users must know the storage location and description of components to search for them, then precise searches are possible, but the system becomes difficult to use when this information is unavailable
Solution Approach 1:
The system performs self-service by automatically analyzing visual features of query components and comparing them against the database without requiring user knowledge of storage locations, component names, or detailed descriptions. The machine learning model autonomously handles the matching process, making the system accessible to users regardless of their familiarity with the data store organization.
Data Source
AI summary
A computing device can receive 3-D model data that includes components included in 3-D models. The components can be represented by 2-D images. The computing device can generate representation vectors that correspond to different 2-D images. The computing device can index the representation vectors to facilitate queries with respect to the components. The computing device can receive a query indicating a request to identify 2-D images associated with the query. The computing device can compare an embedding of the query to the representation vectors to identify a set of representation vectors that match the embedding. The computing device can provide, on a digital user interface, the 2-D images corresponding to the set of representation vectors.