Object functionality prediction method and device, computer equipment and storage medium

Abstract

The invention relates to an object functionality prediction method and device, computer equipment and a storage medium. The method comprises the steps of acquiring a to-be-predicted object and a plurality of candidate scenes; inputting the to-be-predicted object and the current candidate scenes to a distance measurement model, performing calculation by the distance measurement model according to atrained scene feature sub-network to obtain eigenvectors corresponding to the current candidate scenes, and outputting distances from the to-be-predicted object to the current candidate scenes according to the to-be-predicted object and the eigenvectors corresponding to the current candidate scenes, wherein model parameters of the distance measurement model include parameters determined by a trained object feature sub-network; obtaining the distances from the to-be-predicted object to the candidate scenes according to the distance measurement model; determining a target scene corresponding tothe to-be-predicted object according to the distances from the to-be-predicted object to the candidate scenes; and obtaining a functionality prediction result corresponding to the to-be-predicted object according to the target scene. The object functionality prediction universality can be improved.

Description

technical field [0001] The present application relates to the field of computer technology, in particular to a method, device, computer equipment and storage medium for object function prediction. Background technique [0002] Functional analysis of 3D objects has received much attention as a way to understand and manipulate 3D environments. The function of an object can be reflected by the three-dimensional scene containing this object, this object is called the central object, and the central object is surrounded by one or more other objects. [0003] The traditional object function prediction method usually analyzes the object structure in a geometric way to obtain the function, and needs to specify many specific component structure models, which is not universal. Or analyze the interaction between humans and objects by simulating humanoid agents. As for the functions embodied in the interaction between objects, the functional prediction is limited because they cannot be...

AI Technical Summary

Problems solved by technology

Or analyze the interaction between humans and objects by simulating humanoid agents. As for the functions embodied in the interaction between objects, the functional prediction is limited because they cannot be directly analyzed by simulating humanoid agents.

Images