Floor Plan Space Classification via Image Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in efficiently generating attribute data for space management from raw floor plan data, particularly when CAD files are absent or of varying quality, leading to time-consuming manual processing and categorization.
Innovation Solution
A computer system utilizing image classification models, including machine learning models like convolutional neural networks, to classify elements in floor plans, segment spaces, and derive attribute data by processing image data from CAD documents or alternative formats like JPEG and PDF, automatically identifying boundaries and objects to determine space types and capacities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual cleaning, enriching and categorizing of floor plan elements is performed, then data quality and completeness can be improved, but time consumption and processing cost increase significantly
Solution Approach 1:
The patent replaces manual mechanical processing of floor plans with an automated computer-based system that uses image processing and machine learning algorithms to extract, classify, and enrich spatial data from various floor plan formats, thereby eliminating time-consuming manual effort while maintaining or improving data quality
Solution Approach 2:
The system enables floor plan data to be automatically processed and enriched without human intervention by implementing algorithms that can independently identify elements, classify spaces, and generate attribute data from raw floor plan inputs in multiple formats
2Adaptability or versatility
If multiple floor plan formats and sources are supported, then system versatility and adaptability improve, but system complexity and processing difficulty increase
Solution Approach 1:
The patent implements a universal processing framework that can handle multiple floor plan formats (CAD, PDF, images, SVG) through a single integrated system using image processing techniques and machine learning models that automatically adapt to different input formats without requiring separate processing pipelines for each format
Solution Approach 2:
The system introduces an intermediary image processing layer that converts various floor plan formats into a standardized representation suitable for machine learning analysis, thereby simplifying the handling of diverse formats and reducing system complexity
3Productivity
If automated image classification models are used to classify floor plan elements, then processing speed and productivity improve, but classification accuracy and precision may be compromised
Solution Approach 1:
The patent applies preliminary image processing steps including noise reduction, contrast enhancement, and feature extraction before feeding images to classification models, which prepares the data in advance to improve classification accuracy while maintaining automated processing speed
Solution Approach 2:
The system implements feedback mechanisms where classification results are validated and refined through iterative processing, allowing the model to learn from its predictions and improve accuracy over time while maintaining high processing throughput
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The data representing the floor plan is processed to classify the different elements in the floor plan. A group of elements belonging to a subset of the classifications are then selected and used to segment the floor plan into different spaces. For example, certain boundary elements (e.g. the walls and doors) between different spaces (i.e. rooms) in the floor plan may be identified and selected. Once identified, the boundary elements may be used to segment the floor plan into different spaces, which can then be analysed to determine the characteristics of those spaces, in particular the classification of the space into a type (e.g. an intended use for the space or a type of space, such as an, office, kitchen, etc). To perform the classification of the spaces, image data for each space is provided to one or more image classification models, to obtain an indication of the classification.