GeoAIR-lab / TGGLinesPlus

Jupyter notebooks for TGGLinesPlus computer vision algorithm for line detection from images

About TGGLinesPlus

TGGLinesPlus is a free, open-sourced Python implementation of the TGGLinesPlus algorithm for line detection from images released by the GeoAIR Lab (https://geoair.lipingyang.org/). Many problems can be moved forward through TGGLinesPlus, including but not limited to the following: advanced optical character recognition (OCR) techniques, road lane line extraction for real-time autonomous driving, concrete cracks analysis in structural engineering, AutoCAD map vectorization, contour map digitalization, medical image processing and analysis, in addition to feature extraction for machine learning algorithms.

TGGLinesPlus paper

TGGLinesPlus algorithm is detailed in the paper, titled "TGGLinesPlus: A robust topological graph-guided computer vision algorithm for line detection from images".

Copyright

TGGLinesPlus Python implementation is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version. See the file COPYING in this directory or http://www.gnu.org/licenses/, for a description of the GNU General Public License terms under which you can copy the files.

Files

• .gitignore
  Globally ignored files by git

• environment.yml
conda environment description of relevant dependencies

To recreate the conda environment we use in this repository, please run:

conda env create -f environment.yml

And to activate the environment:

conda activate skeleton_graph

If you should ever want to export the full list of package dependencies from this conda environment, you can run:

conda env export --no-builds | grep -v "prefix" > environment.yml

• --no-builds: this is an attempt to make this miniconda environment work on across different operating systems by removing the build information for each package. To read more about this and how to fix any issues you run into if the above commands for installing the environment don't work for you, please see this excellent post
• grep -v "prefix": this hides the prefix portion of the YAML file that lists the directory for where your miniconda environments are housed. This is likely to be different from one computer to the next, so it is not necessary, and hides full path information if you are working in data-sensitive environments.

Folders

data

• mnist: the original MNIST dataset stored in CSV format and a zip file of the Chinese MNIST dataset
• scikit-image the [page] and [retina] ndarrays
• rs_imagery: remote sensing image files
• deepcrack: image 11215-5.png selected from the [DeepCrack] dataset
• mass_roads: image 11278840_15.tif selected from the [Massachusetts Roads] dataset
• contours: [city contours] from Albuquerque, NM; the original shapefile was overlaid onto a basemap of the area using QGIS, then exported as a PNG file

notebooks

All Jupyter Notebook demos are located here, alongside all of the processing and plotting methods used in a folder called utils/.