keras-spatial 1.0.0

Keras Data Generator class for remote spatial data.

Keras Spatial

Keras Spatial includes data generators and utilities designed to simplify working with spatial data.

Keras Spatial provides a data generator based on a raster file. The generator reads directly from this source and eliminates the need to create small raster files in heirarchical directory structure. The raster file may reside locally or remotely. Any necessary reprojections and scaling is handled automatically.

Additional gridding utilties are available to assist in the creation of a dataframes based on the spatial extent of a raster and number of desired samples.

Installation

To install the package from PyPi repository you can execute the following command:

pip install keras-spatial

or directly from GitHub

$ pip install git+https://github.com/IllinoisStateGeologicalSurvey/keras-spatial#egg=keras-spatial --process-dependency-links

Quickstart

1. Create a SpatialDataGen that will create 200, 200 pixel samples from TIF
2. Create a 10,000 element geodataframe covering the spatial extent of the raster
3. Create the generator
4. Fit model

from keras_spatial.datagen import SpatialDataGenerator

sdg = SpatialDataGenerator(200,200, source='/path/to/file.tif')
geodataframe = sdg.regular_grid(.01, .01)
generator = sdg.flow_from_dataframe(geodataframe)
model.fit_generator(generator, ...)

Usage

Keras Spatial provides a SpatialDataGenerator (SDG) modeled on the Keras ImageDataGenerator. The SDG allows user to work in spatial coorindates rather than pixels and easily integrate data from different coordinates systems. Reprojection and resampling is handled automatically as needed. Because Keras Spatial is based on the rasterio package, raster data source may either local files or remote resources referenced by URL.

Because the SDG reads directly from larger raster data sources rather than small, preprocessed images files, SDG makes use of a GeoDataFrame to identify each sample area. The geometry associated with the datafame is expected to be a polygon but extraction is done using a windowed read based on the bounds. As with the ImageDataGenerator, the flow_from_dataframe method returns the generator that can be passed to the Keras model.

SpatialDataGenerator class

The SDG is similar to the ImageDataGenerator albeit missing the .flow and the .flow_from_directory methods. SDG also moves more configutation and setting to the instance and with the .flow_from_dataframe having few arguments.

Arguments

• width (int): sample width in pixels
• height (int): sample height in pixels
• source (path or url): raster source
• indexes (int or sequence of ints): one or more raster bands to sampled
• crs (CRS): the desired coordinate reference system if different from source
• interleave (str): type of interleave 'band' or 'pixel'
• resampling (int): One of the values from rasterio.enums.Resampling (default=Resampling.nearest)
• preprocess (function): callback function invoked on the numpy array prior to returning it to the model

Raises RasterioIOError when the source is set if the file or resource is not available.

Examples

from keras_spatial import SpatialDataGenerator
sdg = SpatialDataGenerator()
sdg.height, sdg.width = 200,200
sdg.source = '/path/to/file.tif'

The height, width, and source arguments are required prior calling flow_from_dataframe. This can be done in several equivalent ways as shown below.

sdg1 = SpatialDataGenerator(200,200)
sdg1.source = '/path/to/file.tif'
sdg2 = SpatialDataGenerator(200,200, source='/path/to/file.tif')

The indexes argument selects bands in a multiband raster. By default all bands are read and the indexes argument is updated when the raster source is set.

If interleave is set to 'pixel' and more than one band is read (based on indexes), the numpy array axes are moved. This can lead to incompatible shapes when using multiple SDG generators -- use with care.

sdg = SpatialDataGenerator(200,200, '/path/to/file.tif')
gen = sdg.flow_from_dataframe(df, batch_size=1)
print(next(gen).shape)
> [1, 5, 200, 200]
sdg.interleave = 'pixel'
gen = sdg.flow_from_dataframe(df, batch_size=1)
print(next(gen.shape))
> [1, 200, 200, 5]

Because more than one SDG is expected to be used simultaneously and SDGs are expected to having matching spatial requirements, the SDG class provides a profile attribute that can be easily share arguments across instances as shown below. Note: source is not part of the profile.

sdg = SpatialDataGenerator(200,200, source='/path/to/file.tif')
sdg2 = SpatialDataGenerator()
sdg2.profile = sdg.profile
sdg2.source = '/path/to/file2.tif'

SpatialDataGenerator methods

flow_from_dataframe

flow_from_dataframe(geodataframe, batch_size)

Creates a generator that returns a numpy ndarray of samples read from the SDG source.

Arguments

• geodataframe (GeoDataFrame): a geodataframe with sample boundaries
• batch_size (int): number of samples to returned by generator

Returns

A generator of numpy ndarrays

Example

sdg = SpatialDataGenerator(200, 200, source='/path/to/file.tif')
gen = sdg.flow_from_dataframe(df)

random_grid

random_grid(pct_width, pct_height, count)

Creates a geodataframe suitable to passing to the flow_from_dataframe method. The sample module provides a similar function using passed using spatial extents.

Arguments

• pct_width (float): percentage of raster width
• pct_height (float): percentage of raster height
• count (int): number of samples

Returns

A GeoDataFrame defining a polygon around each sample.

Example

sdg = SpatialDataGenerator(source='/path/to/file.tif')
df = sdg.random_grid(0.01, 0.01, 1000)

regular_grid

regular_grid(pct_width, pct_height, overlap)

Creates a geodataframe suitable to passing to the flow_from_dataframe method. The sample module provides a similar function using passed using spatial extents.

Arguments

• pct_width (float): percentage of raster width
• pct_height (float): percentage of raster height
• overlap (float): percentage of overlap (default=0.0)

Returns

A GeoDataFrame defining a polygon around each sample.

Example

sdg = SpatialDataGenerator(source='/path/to/file.tif')
df = sdg.regular_grid(0.01, 0.01)

Full Example

from keras_spatial import SpatialDataGenerator
labels = SpatialDataGenerator(200,200)
labels.source = '/path/to/labels.tif'
df = labels.regular_grid(0.01, 0.01)

patches = SpatialDataGenerator(200,200)
patches.source = 'https://server.com/files/data.tif'
patches.crs = labels.crs

train_gen = zip(labels.flow_from_dataframe(df), patches.flow_from_dataframe(df))
model(train_gen)