https://github.com/cansik/architectural-floor-plan

AFPlan is an architectural floor plan analysis and recognition system to create extended plans for building services.
https://github.com/cansik/architectural-floor-plan

dwg floorplans image-classification polygon recognition

Last synced: about 5 hours ago
JSON representation

AFPlan is an architectural floor plan analysis and recognition system to create extended plans for building services.

• Host: GitHub
• URL: https://github.com/cansik/architectural-floor-plan
• Owner: cansik
• Created: 2016-09-28T07:10:46.000Z (about 8 years ago)
• Default Branch: master
• Last Pushed: 2021-10-29T09:42:40.000Z (almost 3 years ago)
• Last Synced: 2023-10-20T18:40:11.444Z (12 months ago)
• Topics: dwg, floorplans, image-classification, polygon, recognition
• Language: Kotlin
• Size: 209 MB
• Stars: 266
• Watchers: 18
• Forks: 67
• Open Issues: 4
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Automatic analysis and simplification of architectural floor plans [![Build Status](https://travis-ci.org/cansik/architectural-floor-plan.svg?branch=master)](https://travis-ci.org/cansik/architectural-floor-plan) [![codebeat badge](https://codebeat.co/badges/244f2179-f84e-4a39-8943-3285d0cf8337)](https://codebeat.co/projects/github-com-cansik-architectural-floor-plan) [![DOI](https://zenodo.org/badge/69436030.svg)](https://zenodo.org/badge/latestdoi/69436030)

This software is an architectural floor plan analysis and recognition system to create extended plans for building services.

## Abstract

The goal of this work is to do a fast and robust room detection on floor plans. The idea is, that a wide range of non standardized floor plans can be analyzed, time efficient, with little drawbacks in its precision.

The used workflow consists of several algorithms, that are combined to deliver the expected result. It consists of *Morphological cleaning* for noise removal, *Machine Learning* and *Convex Hull closing* for gap closing and a *Connected Component analysis* for the final room detection. It is the best result out of different approaches that were tested. All of the algorithms used, use an image of a plan  as the start for detection and return the location and size of each room as a CSV-table or SVG-vectors. The software is prepared to return the rooms as a DWG- or DXF-Format for a CAD-Program, but the license for a library, to convert the format, is not finally evaluated. The algorithm implemented, shows improvement in room detection accuracy, compared to similar works done in the last few years.

![Afpars](readme/afpars.jpg)

## Run the application

The software is still a prototype and not packaged into an executable. To run the software, you have to run the following command:

```bash

cd afpars

# unix / macOS

./gradlew run

# windows

gradlew.bat run

```

## Development prerequisites

For development on the existing project, you have to install a modern Java Development Kit (`>=11`). OpenJDK is supported!

The project itself can be built with the [gradle build tool](https://gradle.org/).

```bash

cd afpars

# unix / macOS

./gradlew build

# windows

gradlew.bat build

```

## About

*FHNW Bachelor Computer Science*

*Alexander Wyss and Florian Bruggisser*