Skip to main content

A python library for preprocessing geospatial vector geometries for use in deep learning

Project description

deep-geometry

A python library for preprocessing geospatial vector geometries for use in deep learning

Rationale

Deep learning can use geospatial vector polygons directly (rather than a feature-extracted pre-processd version), but it requires vectorization and normalisation first, like any data source.

Installation

pip install deep-geometry

Usage

Geometry vectorization

Make a numerical vector from a geometry:

>>> from deep_geometry import vectorizer as gv

>>> geoms = [
...     'POINT(0 0)',
...     'POINT(1 1)',
...     'POINT(2 2)',
...     'POINT(3 3)',
...     'POINT(4 4)',
...     'POINT(5 5)',
...     'POLYGON((0 0, 1 0, 1 1, 0 1, 0 0))',
... ]

>>> gv.vectorize_wkt(geoms[0])
array([[ 0.,  0.,  0.,  0.,  0.,  0.,  1.]])

>>> gv.vectorize_wkt(geoms[6])
array([[ 0.,  0.,  0., 1., 1.,  0.,  0.],
       [ 1.,  0.,  0., 1., 1.,  0.,  0.],
       [ 1.,  1.,  0., 1., 1.,  0.,  0.],
       [ 0.,  1.,  0., 1., 1.,  0.,  0.],
       [ 0.,  0.,  0., 1., 0.,  0.,  1.]])

Collect the max length from a set of geometries:

>>> max_len = gv.get_max_points(geoms)
>>> print('Maximum geometry node size in set:', max_len)
Maximum geometry node size in set: 7

Numerical data normalization

Geometries regularly are in some kind of earth projection that is far from the origin of the coordinate system. In order for machine learning models to learn, data needs to be normalized. A usual way to go about this is to mean-center the instances and to divide by the dataset standard deviation.

The library provides a convenience class for normalization, modeled after the scalers from scikit-learn with a .fit() and a .transform() method:

>>> from deep_geometry import GeomScaler
>>> import numpy
>>> gs = GeomScaler()  # simply initialize
>>> geom6 = gv.vectorize_wkt(geoms[6])
>>> dataset = numpy.repeat([geom6], 4, axis=0)
>>> gs.fit(dataset)
>>> gs.scale_factor
0.5
>>> normalized_data = gs.transform(dataset)
>>> normalized_data[0]  # see: zero-mean and scaled to standard deviation
array([[-1., -1.,  0.,  1.,  1.,  0.,  0.],
       [ 1., -1.,  0.,  1.,  1.,  0.,  0.],
       [ 1.,  1.,  0.,  1.,  1.,  0.,  0.],
       [-1.,  1.,  0.,  1.,  1.,  0.,  0.],
       [-1., -1.,  0.,  1.,  0.,  0.,  1.]])

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for deep-geometry, version 2.0.0
Filename, size File type Python version Upload date Hashes
Filename, size deep_geometry-2.0.0-py3-none-any.whl (10.1 kB) File type Wheel Python version py3 Upload date Hashes View hashes
Filename, size deep-geometry-2.0.0.tar.gz (7.8 kB) File type Source Python version None Upload date Hashes View hashes

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page