ANN Floor Plan Estimation From Noisy 3D Scans
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for generating accurate floor plans from 3D meshes or point clouds are limited by the need for professional skills and time, and fail to handle imperfect scans with unusual angles, object clutter, or open-loop scans, leading to inaccurate or incomplete results.
Innovation Solution
Employing machine learning techniques, specifically artificial neural networks (ANNs) trained with a diverse dataset of top-down views of 3D meshes or point clouds and corresponding ground truth floor plans, to estimate accurate floor plans even in challenging scenarios.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional algorithms are used to generate floor plans from 3D meshes, then the process requires professional skills and significant time, but the results may be inaccurate when dealing with imperfect scans
Solution Approach 1:
The patent replaces traditional manual or algorithmic processing methods with a machine learning-based neural network system. The neural network is trained on diverse datasets including imperfect scans and then automatically generates floor plans, substituting the need for professional manual intervention and complex traditional algorithms while maintaining or improving accuracy.
Solution Approach 2:
The patent employs a preliminary training phase where the neural network is trained on a diverse dataset containing various scan qualities and scenarios before actual floor plan generation. This preliminary action prepares the system to handle imperfect scans and diverse scenarios efficiently during deployment, reducing processing time for actual tasks.
2Adaptability or versatility
If traditional methods are used to handle imperfect scans with unusual angles or object clutter, then the process can be completed, but the results become inaccurate or incomplete
Solution Approach 1:
The patent changes the approach from rule-based traditional algorithms to a data-driven neural network that learns from diverse examples. By training on varied datasets including scans with unusual angles and object clutter, the system adapts its parameters and decision-making processes to handle these challenging scenarios accurately.
Solution Approach 2:
The patent uses a diverse training dataset that includes copies or representations of various scan scenarios (perfect scans, imperfect scans, scans with unusual angles, scans with object clutter). The neural network learns from these copied examples to generalize and accurately process real-world diverse scan inputs.
3Measurement precision
If professional manual methods are used to create accurate floor plans, then high precision is achieved, but the productivity is low
Solution Approach 1:
The patent replaces manual professional methods with an automated neural network system. The neural network, once trained, can generate floor plans automatically without requiring professional manual intervention, thereby dramatically increasing productivity while maintaining the accuracy that professionals previously provided.
Solution Approach 2:
The patent implements a self-service system where the neural network autonomously processes scan data and generates floor plans without requiring manual professional intervention. The system serves itself by automatically handling the entire workflow from input scan to output floor plan, significantly improving productivity.
Data Source
AI summary
Artificial neural networks (ANN) may be trained to output estimated floor plans from 3D spaces that would be challenging or impossible for existing techniques to estimate. In embodiments, an ANN may be trained using a supervised approach where top-down views of 3D meshes or point clouds are provided to the ANN as input, with ground truth floor plans provided as output for comparison. A suitably large training set may be used to fully train the ANN on challenging scenarios such as open loop scans and/or unusual geometries. The trained ANN may then be used to accurately estimate floor plans for such 3D spaces. Other embodiments are described.