Object Functionality Prediction via Feature Distance Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional object functionality prediction methods are limited by their geometric analysis approach, which is not universal and fails to directly analyze the interaction between objects, restricting the prediction of object functions.
Innovation Solution
A method using a computer device to predict object functionality by constructing a function similarity network with trained scene and object feature subnetworks, calculating distances between objects and candidate scenes, and determining a target scene for functionality prediction, allowing for the inference of interaction contexts without relying on geometric structures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional geometric analysis methods are used to predict object functionality, then the method can analyze object structure, but it requires specifying many specific component structure models and is not universal
Solution Approach 1:
The patent replaces traditional geometric analysis methods with a deep learning-based neural network system. The neural network learns functional representations directly from image data without requiring explicit geometric modeling or component structure specifications, thereby achieving universality while reducing complexity
Solution Approach 2:
The patent transforms the prediction approach from geometric parameters to learned feature representations. By using neural networks to automatically extract and compare functional features from images, the system changes the parameter space from explicit geometric measurements to implicit learned representations, enabling universal applicability
2Measurement precision
If humanoid agent simulation is used to analyze human-object interaction, then interaction can be simulated, but functionality prediction is limited due to inability to directly analyze object functions
Solution Approach 1:
The patent creates a universal functionality representation that works across different objects and interaction contexts. The neural network learns a common functional space that can represent both object properties and interaction characteristics, making the system versatile for analyzing various human-object interactions without requiring separate simulation models
Solution Approach 2:
The patent introduces a neural network-based functional representation as an intermediary between image data and functionality prediction. This intermediary transforms visual information into functional descriptors that capture both object properties and interaction potential, enabling direct functionality analysis without humanoid agent simulation
Data Source
AI summary
A method is disclosed. The method includes obtaining an object for prediction and a plurality of candidate scenes by a computer device; inputting the object for prediction and a current candidate scene to a distance measurement model, the distance measurement model calculates a feature vector corresponding to the current candidate scene based on a trained scene feature subnetwork, and outputs a distance from the object for prediction to the current candidate scene based on the object for prediction and the feature vector corresponding to the current candidate scene, model parameters of the distance measurement model including a parameter determined by a trained object feature subnetwork; obtaining distances from the object for prediction to the plurality of candidate scenes based on the distance measurement model; determining a target scene corresponding to the object for prediction based on the distances from the object for prediction to the plurality of candidate scenes.