Mind the Gap is an algorithm for detecting gaps in building footprints datasets, with a preference for finding the rectangular-shaped gaps caused by missing imagery tiles. It works best on country-sized datasets or smaller, or larger datasets divided into smaller chunks.
A Dockerfile and requirements are included, and can help you set up a suitable container.
Alternatively, you can use poetry by running:
pip install poetry
and then
poetry install
from the root directory of the repository.
The core of the algorithm is found in mind_the_gap.py, and you can use mind_the_gap function to directly run the algorith. For an easier start, you can use the auto_tune module, which will free you from having to guess and check all the parameters. This will do a decent job of tuning all the paramters, but you may still need to tweak things. The default parameters in the Region.run method work generally well but may need a little tweaking. Region.run_parallel is advantageous for large datasets as it allows for parallel processing.
Mind the Gap requires two inputs: building footprints (or just their centroids) and a boundary to the aoi. These are best handled using GeoPandas
Gaps will be stored as a GeoDataFrame of the Region object once Region.run has ran.