Average Path Length Similarity (APLS) Plugin for QGIS Software

The APLS Plugin for QGIS Software is a tool that can be used to evaluate extracted road networks from satellite imagery. This plugin uses the Average Path Length Similarity (APLS) matrix to compare the extracted road network to a ground truth network. APLS is a measure of the similarity between two networks, based on the average shortest path length between pairs of nodes.

Requrements

• QGIS 3.0 or higher

Usage

• Before openning the plugin make sure Python console is opened
• Run the plugin and specify inputs correctly
  ex:Ground truth: /main_folder/Predicted_shp/ Predicted : /main_folder/test_shp/ Djkstra Tolerance : 0.0 Snapping Tolerance : 8.0
• Run and wait untill processing is finished
• The results are displayed in Python console Results example :

Processing total of : 49 images.
Apls for Folder : 18178780_15  is  0.9886363636363636
Apls for Folder : 11278840_15  is  0.11675731021559
Apls for Folder : 23278915_15  is  0.3240173376386126
Apls for Folder : 18478900_15  is  0.9666666666666667
................
Apls for Folder : 24628885_15  is  0.07408254856578367
Apls for Folder : 22228900_15  is  0.4204545454545454
Apls for Folder : 24479215_15  is  0.1971564400253033
AVG APLS is 0.4584244128431006

Inputs

• Ground truth folder contains sub-folders each sub-folder contain shapefile
• Predicted network folder which contains sub-folders each sub-folder contain shapefile
• Djkstra Algorithm Tolerance
• Snapping Tolerance

The plugin requires QGIS 3.0 or higher to be installed and allows users to input a ground truth folder containing sub-folders, each of which contains a shapefile, and a predicted network folder containing sub-folders, each of which also contains a shapefile. The plugin then calculates the APLS for each sub-folder in the predicted network folder and outputs the results in the Python console. Example of Folder Structure for the files

/main_folder
│   └── test_shp/
│       ├── 10378780_15
│       │   ├── 10378780_15.cpg
│       │   ├── 10378780_15.dbf
│       │   ├── 10378780_15.prj
│       │   ├── 10378780_15.shp
│       │   ├── 10378780_15.shx
│       ├── 10828720_15
│       │   ├── 10828720_15.cpg
│       │   ├── 10828720_15.dbf
│       │   ├── 10828720_15.prj
│       │   ├── 10828720_15.shp
│       │   ├── 10828720_15.shx
-----------------------------------
│   └── Predicted_shp/
│       ├── 10378780_15
│       │   ├── 10378780_15.cpg
│       │   ├── 10378780_15.dbf
│       │   ├── 10378780_15.prj
│       │   ├── 10378780_15.shp
│       │   ├── 10378780_15.shx
│       ├── 10828720_15
│       │   ├── 10828720_15.cpg
│       │   ├── 10828720_15.dbf
│       │   ├── 10828720_15.prj
│       │   ├── 10828720_15.shp
│       │   ├── 10828720_15.shx

Screenshot

Resources

@inproceedings{van2020road,
  title={Road network and travel time extraction from multiple look angles with spacenet data},
  author={Van Etten, Adam and Shermeyer, Jacob and Hogan, Daniel and Weir, Nicholas and Lewis, Ryan},
  booktitle={IGARSS 2020-2020 IEEE International Geoscience and Remote Sensing Symposium},
  pages={3920--3923},
  year={2020},
  organization={IEEE}
}

Citation

@masterthesis{MohammedThesisRoad,
    author = {Mohammed Nasser},
    title = {Road Identification from Satellite Imagery Using Deep Learning},
    school = {Erciyes University},
    year = {2022}
}