3D Object Recognition via 2D Segmentation Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual audits of physical infrastructure assets are prone to errors and pose safety risks due to the need for on-site inspections, especially in remote or hazardous locations, and existing methods lack efficient three-dimensional object recognition capabilities.
Innovation Solution
A method using electronic processing devices to determine two-dimensional images, segmentations, and generate three-dimensional representations, mapping these segmentations to recognize objects in a scene, employing techniques like photogrammetry and convolution neural networks for accurate and remote infrastructure assessment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual audits are conducted by persons physically attending infrastructure sites, then asset inspection and documentation can be performed, but safety risks increase and measurement accuracy may be compromised due to dangerous conditions
Solution Approach 1:
The patent creates a three-dimensional digital copy (reality model) of the infrastructure asset from multiple two-dimensional images. This digital replica allows detailed inspection and measurement without requiring personnel to physically access hazardous locations, thereby eliminating safety risks while maintaining measurement accuracy through precise image processing and 3D reconstruction algorithms
Solution Approach 2:
The patent replaces manual physical inspection with an automated computer vision system that processes images and performs measurements algorithmically. This substitution of mechanical/manual operations with automated digital processing eliminates human exposure to dangerous conditions while improving measurement consistency and accuracy through systematic image analysis
2Measurement precision
If three-dimensional object recognition is implemented using multiple two-dimensional images, then accurate remote infrastructure assessment is achieved, but computational intensity increases
Solution Approach 1:
The patent performs preliminary processing by generating the three-dimensional reality model and mapping structures before conducting detailed object recognition and measurement. This pre-computation organizes the data in advance, creating an efficient framework that reduces the computational burden during actual analysis phases while maintaining high measurement precision
Solution Approach 2:
The patent divides the complex three-dimensional recognition task into separate two-dimensional image processing steps. By processing images individually, generating segmentations for each, and then mapping them to the 3D model, the system breaks down computational complexity into manageable segments that can be processed more efficiently while achieving accurate object recognition
Data Source
AI summary
The present application relates to a method for recognising at least one object in a three-dimensional scene, the method including, in an electronic processing device: determining a plurality of two-dimensional images of the scene, the images at least partially including the at least one object; determining a plurality of two-dimensional segmentations of the at least one object, the two-dimensional segmentations corresponding to the two dimensional images; generating a three-dimensional representation of the scene using the images; generating a mapping indicative of a correspondence between the images and the representation; and using the mapping to map the plurality of segmentations to the three dimensional representation, to thereby recognise the at least one object in the scene.