Shortcuts

torchgeo.datasets

In torchgeo, we define two types of datasets: Geospatial Datasets and Non-geospatial Datasets. These abstract base classes are documented in more detail in Base Classes.

Geospatial Datasets

GeoDataset is designed for datasets that contain geospatial information, like latitude, longitude, coordinate system, and projection. Datasets containing this kind of information can be combined using IntersectionDataset and UnionDataset.

Canadian Building Footprints

class torchgeo.datasets.CanadianBuildingFootprints(root='data', crs=None, res=1e-05, transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VectorDataset

Canadian Building Footprints dataset.

The Canadian Building Footprints dataset contains 11,842,186 computer generated building footprints in all Canadian provinces and territories in GeoJSON format. This data is freely available for download and use.

__init__(root='data', crs=None, res=1e-05, transforms=None, download=False, checksum=False)[source]

Initialize a new Dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • crs (Optional[rasterio.crs.CRS]) – coordinate reference system (CRS) to warp to (defaults to the CRS of the first file found)

  • res (float) – resolution of the dataset in units of CRS

  • transforms (Optional[Callable[[Dict[str, Any]], Dict[str, Any]]]) – a function/transform that takes an input sample and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises
  • FileNotFoundError – if no files are found in root

  • RuntimeError – if download=False and data is not found, or checksum=True and checksums don’t match

Chesapeake Bay High-Resolution Land Cover Project

class torchgeo.datasets.Chesapeake(root='data', crs=None, res=None, transforms=None, cache=True, download=False, checksum=False)

Bases: torchgeo.datasets.RasterDataset, abc.ABC

Abstract base class for all Chesapeake datasets.

Chesapeake Bay High-Resolution Land Cover Project dataset.

This dataset was collected by the Chesapeake Conservancy’s Conservation Innovation Center (CIC) in partnership with the University of Vermont and WorldView Solutions, Inc. It consists of one-meter resolution land cover information for the Chesapeake Bay watershed (~100,000 square miles of land).

For more information, see:

__init__(root='data', crs=None, res=None, transforms=None, cache=True, download=False, checksum=False)[source]

Initialize a new Dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • crs (Optional[rasterio.crs.CRS]) – coordinate reference system (CRS) to warp to (defaults to the CRS of the first file found)

  • res (Optional[float]) – resolution of the dataset in units of CRS (defaults to the resolution of the first file found)

  • transforms (Optional[Callable[[Dict[str, Any]], Dict[str, Any]]]) – a function/transform that takes an input sample and returns a transformed version

  • cache (bool) – if True, cache file handle to speed up repeated sampling

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises
abstract property base_folder: str

Parent directory of dataset in URL.

abstract property filename: str

Filename to find/store dataset in.

abstract property md5: str

MD5 checksum to verify integrity of dataset.

property url: str

URL to download dataset from.

abstract property zipfile: str

Name of zipfile in download URL.

class torchgeo.datasets.Chesapeake7(root='data', crs=None, res=None, transforms=None, cache=True, download=False, checksum=False)

Bases: torchgeo.datasets.Chesapeake

Complete 7-class dataset.

This version of the dataset is composed of 7 classes:

  1. No Data: Background values

  2. Water: All areas of open water including ponds, rivers, and lakes

  3. Tree Canopy and Shrubs: All woody vegetation including trees and shrubs

  4. Low Vegetation: Plant material less than 2 meters in height including lawns

  5. Barren: Areas devoid of vegetation consisting of natural earthen material

  6. Impervious Surfaces: Human-constructed surfaces less than 2 meters in height

  7. Impervious Roads: Impervious surfaces that are used for transportation

  8. Aberdeen Proving Ground: U.S. Army facility with no labels

class torchgeo.datasets.Chesapeake13(root='data', crs=None, res=None, transforms=None, cache=True, download=False, checksum=False)

Bases: torchgeo.datasets.Chesapeake

Complete 13-class dataset.

This version of the dataset is composed of 13 classes:

  1. No Data: Background values

  2. Water: All areas of open water including ponds, rivers, and lakes

  3. Wetlands: Low vegetation areas located along marine or estuarine regions

  4. Tree Canopy: Deciduous and evergreen woody vegetation over 3-5 meters in height

  5. Shrubland: Heterogeneous woody vegetation including shrubs and young trees

  6. Low Vegetation: Plant material less than 2 meters in height including lawns

  7. Barren: Areas devoid of vegetation consisting of natural earthen material

  8. Structures: Human-constructed objects made of impervious materials

  9. Impervious Surfaces: Human-constructed surfaces less than 2 meters in height

  10. Impervious Roads: Impervious surfaces that are used for transportation

  11. Tree Canopy over Structures: Tree cover overlapping impervious structures

  12. Tree Canopy over Impervious Surfaces: Tree cover overlapping impervious surfaces

  13. Tree Canopy over Impervious Roads: Tree cover overlapping impervious roads

  14. Aberdeen Proving Ground: U.S. Army facility with no labels

class torchgeo.datasets.ChesapeakeDC(root='data', crs=None, res=None, transforms=None, cache=True, download=False, checksum=False)

Bases: torchgeo.datasets.Chesapeake

This subset of the dataset contains data only for Washington, D.C.

class torchgeo.datasets.ChesapeakeDE(root='data', crs=None, res=None, transforms=None, cache=True, download=False, checksum=False)

Bases: torchgeo.datasets.Chesapeake

This subset of the dataset contains data only for Delaware.

class torchgeo.datasets.ChesapeakeMD(root='data', crs=None, res=None, transforms=None, cache=True, download=False, checksum=False)

Bases: torchgeo.datasets.Chesapeake

This subset of the dataset contains data only for Maryland.

Note

This dataset requires the following additional library to be installed:

class torchgeo.datasets.ChesapeakeNY(root='data', crs=None, res=None, transforms=None, cache=True, download=False, checksum=False)

Bases: torchgeo.datasets.Chesapeake

This subset of the dataset contains data only for New York.

Note

This dataset requires the following additional library to be installed:

class torchgeo.datasets.ChesapeakePA(root='data', crs=None, res=None, transforms=None, cache=True, download=False, checksum=False)

Bases: torchgeo.datasets.Chesapeake

This subset of the dataset contains data only for Pennsylvania.

class torchgeo.datasets.ChesapeakeVA(root='data', crs=None, res=None, transforms=None, cache=True, download=False, checksum=False)

Bases: torchgeo.datasets.Chesapeake

This subset of the dataset contains data only for Virginia.

Note

This dataset requires the following additional library to be installed:

class torchgeo.datasets.ChesapeakeWV(root='data', crs=None, res=None, transforms=None, cache=True, download=False, checksum=False)

Bases: torchgeo.datasets.Chesapeake

This subset of the dataset contains data only for West Virginia.

class torchgeo.datasets.ChesapeakeCVPR(root='data', splits=['de-train'], layers=['naip-new', 'lc'], transforms=None, cache=True, download=False, checksum=False)

Bases: torchgeo.datasets.GeoDataset

CVPR 2019 Chesapeake Land Cover dataset.

The CVPR 2019 Chesapeake Land Cover dataset contains two layers of NAIP aerial imagery, Landsat 8 leaf-on and leaf-off imagery, Chesapeake Bay land cover labels, NLCD land cover labels, and Microsoft building footprint labels.

This dataset was organized to accompany the 2019 CVPR paper, “Large Scale High-Resolution Land Cover Mapping with Multi-Resolution Data”.

The paper “Resolving label uncertainty with implicit generative models” added an additional layer of data to this dataset containing a prior over the Chesapeake Bay land cover classes generated from the NLCD land cover labels. For more information about this layer see the dataset documentation.

If you use this dataset in your research, please cite the following paper:

__getitem__(query)[source]

Retrieve image/mask and metadata indexed by query.

Parameters

query (torchgeo.datasets.BoundingBox) – (minx, maxx, miny, maxy, mint, maxt) coordinates to index

Returns

sample of image/mask and metadata at that index

Raises

IndexError – if query is not found in the index

Return type

Dict[str, Any]

__init__(root='data', splits=['de-train'], layers=['naip-new', 'lc'], transforms=None, cache=True, download=False, checksum=False)[source]

Initialize a new Dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • splits (Sequence[str]) – a list of strings in the format “{state}-{train,val,test}” indicating the subset of data to use, for example “ny-train”

  • layers (List[str]) – a list containing a subset of “naip-new”, “naip-old”, “lc”, “nlcd”, “landsat-leaf-on”, “landsat-leaf-off”, “buildings”, or “prior_from_cooccurrences_101_31_no_osm_no_buildings” indicating which layers to load

  • transforms (Optional[Callable[[Dict[str, Any]], Dict[str, Any]]]) – a function/transform that takes an input sample and returns a transformed version

  • cache (bool) – if True, cache file handle to speed up repeated sampling

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises

Cropland Data Layer (CDL)

class torchgeo.datasets.CDL(root='data', crs=None, res=None, transforms=None, cache=True, download=False, checksum=False)

Bases: torchgeo.datasets.RasterDataset

Cropland Data Layer (CDL) dataset.

The Cropland Data Layer, hosted on CropScape, provides a raster, geo-referenced, crop-specific land cover map for the continental United States. The CDL also includes a crop mask layer and planting frequency layers, as well as boundary, water and road layers. The Boundary Layer options provided are County, Agricultural Statistics Districts (ASD), State, and Region. The data is created annually using moderate resolution satellite imagery and extensive agricultural ground truth.

If you use this dataset in your research, please cite it using the following format:

__init__(root='data', crs=None, res=None, transforms=None, cache=True, download=False, checksum=False)[source]

Initialize a new Dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • crs (Optional[rasterio.crs.CRS]) – coordinate reference system (CRS) to warp to (defaults to the CRS of the first file found)

  • res (Optional[float]) – resolution of the dataset in units of CRS (defaults to the resolution of the first file found)

  • transforms (Optional[Callable[[Dict[str, Any]], Dict[str, Any]]]) – a function/transform that takes an input sample and returns a transformed version

  • cache (bool) – if True, cache file handle to speed up repeated sampling

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 after downloading files (may be slow)

Raises

Landsat

class torchgeo.datasets.Landsat(root='data', crs=None, res=None, bands=[], transforms=None, cache=True)

Bases: torchgeo.datasets.RasterDataset, abc.ABC

Abstract base class for all Landsat datasets.

Landsat is a joint NASA/USGS program, providing the longest continuous space-based record of Earth’s land in existence.

If you use this dataset in your research, please cite it using the following format:

__init__(root='data', crs=None, res=None, bands=[], transforms=None, cache=True)[source]

Initialize a new Dataset instance.

Parameters
Raises

FileNotFoundError – if no files are found in root

class torchgeo.datasets.Landsat9(root='data', crs=None, res=None, bands=[], transforms=None, cache=True)

Bases: torchgeo.datasets.Landsat8

Landsat 9 Operational Land Imager (OLI) and Thermal Infrared Sensor (TIRS).

class torchgeo.datasets.Landsat8(root='data', crs=None, res=None, bands=[], transforms=None, cache=True)

Bases: torchgeo.datasets.Landsat

Landsat 8 Operational Land Imager (OLI) and Thermal Infrared Sensor (TIRS).

class torchgeo.datasets.Landsat7(root='data', crs=None, res=None, bands=[], transforms=None, cache=True)

Bases: torchgeo.datasets.Landsat

Landsat 7 Enhanced Thematic Mapper Plus (ETM+).

class torchgeo.datasets.Landsat5TM(root='data', crs=None, res=None, bands=[], transforms=None, cache=True)

Bases: torchgeo.datasets.Landsat4TM

Landsat 5 Thematic Mapper (TM).

class torchgeo.datasets.Landsat5MSS(root='data', crs=None, res=None, bands=[], transforms=None, cache=True)

Bases: torchgeo.datasets.Landsat4MSS

Landsat 4 Multispectral Scanner (MSS).

class torchgeo.datasets.Landsat4TM(root='data', crs=None, res=None, bands=[], transforms=None, cache=True)

Bases: torchgeo.datasets.Landsat

Landsat 4 Thematic Mapper (TM).

class torchgeo.datasets.Landsat4MSS(root='data', crs=None, res=None, bands=[], transforms=None, cache=True)

Bases: torchgeo.datasets.Landsat

Landsat 4 Multispectral Scanner (MSS).

class torchgeo.datasets.Landsat3(root='data', crs=None, res=None, bands=[], transforms=None, cache=True)

Bases: torchgeo.datasets.Landsat1

Landsat 3 Multispectral Scanner (MSS).

class torchgeo.datasets.Landsat2(root='data', crs=None, res=None, bands=[], transforms=None, cache=True)

Bases: torchgeo.datasets.Landsat1

Landsat 2 Multispectral Scanner (MSS).

class torchgeo.datasets.Landsat1(root='data', crs=None, res=None, bands=[], transforms=None, cache=True)

Bases: torchgeo.datasets.Landsat

Landsat 1 Multispectral Scanner (MSS).

National Agriculture Imagery Program (NAIP)

class torchgeo.datasets.NAIP(root, crs=None, res=None, transforms=None, cache=True)

Bases: torchgeo.datasets.RasterDataset

National Agriculture Imagery Program (NAIP) dataset.

The National Agriculture Imagery Program (NAIP) acquires aerial imagery during the agricultural growing seasons in the continental U.S. A primary goal of the NAIP program is to make digital ortho photography available to governmental agencies and the public within a year of acquisition.

NAIP is administered by the USDA’s Farm Service Agency (FSA) through the Aerial Photography Field Office in Salt Lake City. This “leaf-on” imagery is used as a base layer for GIS programs in FSA’s County Service Centers, and is used to maintain the Common Land Unit (CLU) boundaries.

If you use this dataset in your research, please cite it using the following format:

Sentinel

class torchgeo.datasets.Sentinel(root, crs=None, res=None, transforms=None, cache=True)

Bases: torchgeo.datasets.RasterDataset

Abstract base class for all Sentinel datasets.

Sentinel is a family of satellites launched by the European Space Agency (ESA) under the Copernicus Programme.

If you use this dataset in your research, please cite it using the following format:

class torchgeo.datasets.Sentinel2(root='data', crs=None, res=None, bands=[], transforms=None, cache=True)

Bases: torchgeo.datasets.Sentinel

Sentinel-2 dataset.

The Copernicus Sentinel-2 mission comprises a constellation of two polar-orbiting satellites placed in the same sun-synchronous orbit, phased at 180° to each other. It aims at monitoring variability in land surface conditions, and its wide swath width (290 km) and high revisit time (10 days at the equator with one satellite, and 5 days with 2 satellites under cloud-free conditions which results in 2-3 days at mid-latitudes) will support monitoring of Earth’s surface changes.

__init__(root='data', crs=None, res=None, bands=[], transforms=None, cache=True)[source]

Initialize a new Dataset instance.

Parameters
Raises

FileNotFoundError – if no files are found in root

Non-geospatial Datasets

VisionDataset is designed for datasets that lack geospatial information. These datasets can still be combined using ConcatDataset.

ADVANCE (AuDio Visual Aerial sceNe reCognition datasEt)

class torchgeo.datasets.ADVANCE(root='data', transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionDataset

ADVANCE dataset.

The ADVANCE dataset is a dataset for audio visual scene recognition.

Dataset features:

  • 5,075 pairs of geotagged audio recordings and images

  • three spectral bands - RGB (512x512 px)

  • 10-second audio recordings

Dataset format:

  • images are three-channel jpgs

  • audio files are in wav format

Dataset classes:

  1. airport

  2. beach

  3. bridge

  4. farmland

  5. forest

  6. grassland

  7. harbour

  8. lake

  9. orchard

  10. residential

  11. sparse shrub land

  12. sports land

  13. train station

If you use this dataset in your research, please cite the following paper:

Note

This dataset requires the following additional library to be installed:

  • scipy to load the audio files to tensors

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', transforms=None, download=False, checksum=False)[source]

Initialize a new ADVANCE dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises

RuntimeError – if download=False and data is not found, or checksums don’t match

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

New in version 0.2.

Smallholder Cashew Plantations in Benin

class torchgeo.datasets.BeninSmallHolderCashews(root='data', chip_size=256, stride=128, bands=('B01', 'B02', 'B03', 'B04', 'B05', 'B06', 'B07', 'B08', 'B8A', 'B09', 'B11', 'B12', 'CLD'), transforms=None, download=False, api_key=None, checksum=False, verbose=False)

Bases: torchgeo.datasets.VisionDataset

Smallholder Cashew Plantations in Benin dataset.

This dataset contains labels for cashew plantations in a 120 km2 area in the center of Benin. Each pixel is classified for Well-managed plantation, Poorly-managed plantation, No plantation and other classes. The labels are generated using a combination of ground data collection with a handheld GPS device, and final corrections based on Airbus Pléiades imagery. See this website for dataset details.

Specifically, the data consists of Sentinel 2 imagery from a 120 km2 area in the center of Benin over 71 points in time from 11/05/2019 to 10/30/2020 and polygon labels for 6 classes:

  1. No data

  2. Well-managed plantation

  3. Poorly-managed planatation

  4. Non-plantation

  5. Residential

  6. Background

  7. Uncertain

If you use this dataset in your research, please cite the following:

Note

This dataset requires the following additional library to be installed:

  • radiant-mlhub to download the imagery and labels from the Radiant Earth MLHub

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

image, mask, and metadata at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', chip_size=256, stride=128, bands=('B01', 'B02', 'B03', 'B04', 'B05', 'B06', 'B07', 'B08', 'B8A', 'B09', 'B11', 'B12', 'CLD'), transforms=None, download=False, api_key=None, checksum=False, verbose=False)[source]

Initialize a new Benin Smallholder Cashew Plantations Dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • chip_size (int) – size of chips

  • stride (int) – spacing between chips, if less than chip_size, then there will be overlap between chips

  • bands (Tuple[str, ...]) – the subset of bands to load

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • api_key (Optional[str]) – a RadiantEarth MLHub API key to use for downloading the dataset

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

  • verbose (bool) – if True, print messages when new tiles are loaded

Raises

RuntimeError – if download=False but dataset is missing or checksum fails

__len__()[source]

Return the number of chips in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, time_step=0, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
  • sample (Dict[str, torch.Tensor]) – a sample returned by __getitem__()

  • show_titles (bool) – flag indicating whether to show titles above each panel

  • time_step (int) – time step at which to access image, beginning with 0

  • suptitle (Optional[str]) – optional string to use as a suptitle

Returns

a matplotlib Figure with the rendered sample

Raises

ValueError – if the RGB bands are not included in self.bands

Return type

matplotlib.figure.Figure

New in version 0.2.

BigEarthNet

class torchgeo.datasets.BigEarthNet(root='data', split='train', bands='all', num_classes=19, transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionDataset

BigEarthNet dataset.

The BigEarthNet dataset is a dataset for multilabel remote sensing image scene classification.

Dataset features:

  • 590,326 patches from 125 Sentinel-1 and Sentinel-2 tiles

  • Imagery from tiles in Europe between Jun 2017 - May 2018

  • 12 spectral bands with 10-60 m per pixel resolution (base 120x120 px)

  • 2 synthetic aperture radar bands (120x120 px)

  • 43 or 19 scene classes from the 2018 CORINE Land Cover database (CLC 2018)

Dataset format:

  • images are composed of multiple single channel geotiffs

  • labels are multiclass, stored in a single json file per image

  • mapping of Sentinel-1 to Sentinel-2 patches are within Sentinel-1 json files

  • Sentinel-1 bands: (VV, VH)

  • Sentinel-2 bands: (B01, B02, B03, B04, B05, B06, B07, B08, B8A, B09, B11, B12)

  • All bands: (VV, VH, B01, B02, B03, B04, B05, B06, B07, B08, B8A, B09, B11, B12)

  • Sentinel-2 bands are of different spatial resolutions and upsampled to 10m

Dataset classes (43):

  1. Agro-forestry areas

  2. Airports

  3. Annual crops associated with permanent crops

  4. Bare rock

  5. Beaches, dunes, sands

  6. Broad-leaved forest

  7. Burnt areas

  8. Coastal lagoons

  9. Complex cultivation patterns

  10. Coniferous forest

  11. Construction sites

  12. Continuous urban fabric

  13. Discontinuous urban fabric

  14. Dump sites

  15. Estuaries

  16. Fruit trees and berry plantations

  17. Green urban areas

  18. Industrial or commercial units

  19. Inland marshes

  20. Intertidal flats

  21. Land principally occupied by agriculture, with significant areas of natural vegetation

  22. Mineral extraction sites

  23. Mixed forest

  24. Moors and heathland

  25. Natural grassland

  26. Non-irrigated arable land

  27. Olive groves

  28. Pastures

  29. Peatbogs

  30. Permanently irrigated land

  31. Port areas

  32. Rice fields

  33. Road and rail networks and associated land

  34. Salines

  35. Salt marshes

  36. Sclerophyllous vegetation

  37. Sea and ocean

  38. Sparsely vegetated areas

  39. Sport and leisure facilities

  40. Transitional woodland/shrub

  41. Vineyards

  42. Water bodies

  43. Water courses

Dataset classes (19):

  1. Urban fabric

  2. Industrial or commercial units

  3. Arable land

  4. Permanent crops

  5. Pastures

  6. Complex cultivation patterns

  7. Land principally occupied by agriculture, with significant areas of natural vegetation

  8. Agro-forestry areas

  9. Broad-leaved forest

  10. Coniferous forest

  11. Mixed forest

  12. Natural grassland and sparsely vegetated areas

  13. Moors, heathland and sclerophyllous vegetation

  14. Transitional woodland, shrub

  15. Beaches, dunes, sands

  16. Inland wetlands

  17. Coastal wetlands

  18. Inland waters

  19. Marine waters

If you use this dataset in your research, please cite the following paper:

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', split='train', bands='all', num_classes=19, transforms=None, download=False, checksum=False)[source]

Initialize a new BigEarthNet dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – train/val/test split to load

  • bands (str) – load Sentinel-1 bands, Sentinel-2, or both. one of {s1, s2, all}

  • num_classes (int) – number of classes to load in target. one of {19, 43}

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Raises

ValueError – if self.bands is “s1”

Return type

matplotlib.figure.Figure

New in version 0.2.

Cars Overhead With Context (COWC)

class torchgeo.datasets.COWC(root='data', split='train', transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionDataset, abc.ABC

Abstract base class for the COWC dataset.

The Cars Overhead With Context (COWC) data set is a large set of annotated cars from overhead. It is useful for training a device such as a deep neural network to learn to detect and/or count cars.

The dataset has the following attributes:

  1. Data from overhead at 15 cm per pixel resolution at ground (all data is EO).

  2. Data from six distinct locations: Toronto, Canada; Selwyn, New Zealand; Potsdam and Vaihingen, Germany; Columbus, Ohio and Utah, United States.

  3. 32,716 unique annotated cars. 58,247 unique negative examples.

  4. Intentional selection of hard negative examples.

  5. Established baseline for detection and counting tasks.

  6. Extra testing scenes for use after validation.

If you use this dataset in your research, please cite the following paper:

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', split='train', transforms=None, download=False, checksum=False)[source]

Initialize a new COWC dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train” or “test”

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises
  • AssertionError – if split argument is invalid

  • RuntimeError – if download=False and data is not found, or checksums don’t match

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

abstract property base_url: str

Base URL to download dataset from.

abstract property filename: str

Filename containing train/test split and target labels.

abstract property filenames: List[str]

List of files to download.

abstract property md5s: List[str]

List of MD5 checksums of files to download.

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

New in version 0.2.

class torchgeo.datasets.COWCCounting(root='data', split='train', transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.COWC

COWC Dataset for car counting.

class torchgeo.datasets.COWCDetection(root='data', split='train', transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.COWC

COWC Dataset for car detection.

CV4A Kenya Crop Type Competition

class torchgeo.datasets.CV4AKenyaCropType(root='data', chip_size=256, stride=128, bands=('B01', 'B02', 'B03', 'B04', 'B05', 'B06', 'B07', 'B08', 'B8A', 'B09', 'B11', 'B12', 'CLD'), transforms=None, download=False, api_key=None, checksum=False, verbose=False)

Bases: torchgeo.datasets.VisionDataset

CV4A Kenya Crop Type dataset.

Used in a competition in the Computer Vision for Agriculture (CV4A) workshop in ICLR 2020. See this website for dataset details.

Consists of 4 tiles of Sentinel 2 imagery from 13 different points in time.

Each tile has:

  • 13 multi-band observations throughout the growing season. Each observation includes 12 bands from Sentinel-2 L2A product, and a cloud probability layer. The twelve bands are [B01, B02, B03, B04, B05, B06, B07, B08, B8A, B09, B11, B12] (refer to Sentinel-2 documentation for more information about the bands). The cloud probability layer is a product of the Sentinel-2 atmospheric correction algorithm (Sen2Cor) and provides an estimated cloud probability (0-100%) per pixel. All of the bands are mapped to a common 10 m spatial resolution grid.

  • A raster layer indicating the crop ID for the fields in the training set.

  • A raster layer indicating field IDs for the fields (both training and test sets). Fields with a crop ID 0 are the test fields.

There are 3,286 fields in the train set and 1,402 fields in the test set.

If you use this dataset in your research, please cite the following paper:

Note

This dataset requires the following additional library to be installed:

  • radiant-mlhub to download the imagery and labels from the Radiant Earth MLHub

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data, labels, field ids, and metadata at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', chip_size=256, stride=128, bands=('B01', 'B02', 'B03', 'B04', 'B05', 'B06', 'B07', 'B08', 'B8A', 'B09', 'B11', 'B12', 'CLD'), transforms=None, download=False, api_key=None, checksum=False, verbose=False)[source]

Initialize a new CV4A Kenya Crop Type Dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • chip_size (int) – size of chips

  • stride (int) – spacing between chips, if less than chip_size, then there will be overlap between chips

  • bands (Tuple[str, ...]) – the subset of bands to load

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • api_key (Optional[str]) – a RadiantEarth MLHub API key to use for downloading the dataset

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

  • verbose (bool) – if True, print messages when new tiles are loaded

Raises

RuntimeError – if download=False but dataset is missing or checksum fails

__len__()[source]

Return the number of chips in the dataset.

Returns

length of the dataset

Return type

int

get_splits()[source]

Get the field_ids for the train/test splits from the dataset directory.

Returns

list of training field_ids and list of testing field_ids

Return type

Tuple[List[int], List[int]]

plot(sample, show_titles=True, time_step=0, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
  • sample (Dict[str, torch.Tensor]) – a sample returned by __getitem__()

  • show_titles (bool) – flag indicating whether to show titles above each panel

  • time_step (int) – time step at which to access image, beginning with 0

  • suptitle (Optional[str]) – optional suptitle to use for figure

Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

New in version 0.2.

ETCI2021 Flood Detection

class torchgeo.datasets.ETCI2021(root='data', split='train', transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionDataset

ETCI 2021 Flood Detection dataset.

The ETCI2021 dataset is a dataset for flood detection

Dataset features:

  • 33,405 VV & VH Sentinel-1 Synthetic Aperture Radar (SAR) images

  • 2 binary masks per image representing water body & flood, respectively

  • 2 polarization band images (VV, VH) of 3 RGB channels per band

  • 3 RGB channels per band generated by the Hybrid Pluggable Processing Pipeline (hyp3)

  • Images with 5x20m per pixel resolution (256x256) px) taken in Interferometric Wide Swath acquisition mode

  • Flood events from 5 different regions

Dataset format:

  • VV band three-channel png

  • VH band three-channel png

  • water body mask single-channel png where no water body = 0, water body = 255

  • flood mask single-channel png where no flood = 0, flood = 255

Dataset classes:

  1. no flood/water

  2. flood/water

If you use this dataset in your research, please add the following to your acknowledgements section:

The authors would like to thank the NASA Earth Science Data Systems Program,
NASA Digital Transformation AI/ML thrust, and IEEE GRSS for organizing
the ETCI competition.
__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', split='train', transforms=None, download=False, checksum=False)[source]

Initialize a new ETCI 2021 dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train”, “val”, or “test”

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises
  • AssertionError – if split argument is invalid

  • RuntimeError – if download=False and data is not found, or checksums don’t match

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

EuroSAT

class torchgeo.datasets.EuroSAT(root='data', split='train', transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionClassificationDataset

EuroSAT dataset.

The EuroSAT dataset is based on Sentinel-2 satellite images covering 13 spectral bands and consists of 10 target classes with a total of 27,000 labeled and geo-referenced images.

Dataset format:

  • rasters are 13-channel GeoTiffs

  • labels are values in the range [0,9]

Dataset classes:

  • Industrial Buildings

  • Residential Buildings

  • Annual Crop

  • Permanent Crop

  • River

  • Sea and Lake

  • Herbaceous Vegetation

  • Highway

  • Pasture

  • Forest

This dataset uses the train/val/test splits defined in the “In-domain representation learning for remote sensing” paper:

If you use this dataset in your research, please cite the following papers:

__init__(root='data', split='train', transforms=None, download=False, checksum=False)[source]

Initialize a new EuroSAT dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train”, “val”, or “test”

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises

RuntimeError – if download=False and data is not found, or checksums don’t match

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

New in version 0.2.

FAIR1M (Fine-grAined object recognItion in high-Resolution imagery)

class torchgeo.datasets.FAIR1M(root='data', transforms=None, checksum=False)

Bases: torchgeo.datasets.VisionDataset

FAIR1M dataset.

The FAIR1M dataset is a dataset for remote sensing fine-grained oriented object detection.

Dataset features:

  • 15,000+ images with 0.3-0.8 m per pixel resolution (1,000-10,000 px)

  • 1 million object instances

  • 5 object categories, 37 object sub-categories

  • three spectral bands - RGB

  • images taken by Gaofen satellites and Google Earth

Dataset format:

  • images are three-channel tiffs

  • labels are xml files with PASCAL VOC like annotations

Dataset classes:

  1. Passenger Ship

  2. Motorboat

  3. Fishing Boat

  4. Tugboat

  5. other-ship

  6. Engineering Ship

  7. Liquid Cargo Ship

  8. Dry Cargo Ship

  9. Warship

  10. Small Car

  11. Bus

  12. Cargo Truck

  13. Dump Truck

  14. other-vehicle

  15. Van

  16. Trailer

  17. Tractor

  18. Excavator

  19. Truck Tractor

  20. Boeing737

  21. Boeing747

  22. Boeing777

  23. Boeing787

  24. ARJ21

  25. C919

  26. A220

  27. A321

  28. A330

  29. A350

  30. other-airplane

  31. Baseball Field

  32. Basketball Court

  33. Football Field

  34. Tennis Court

  35. Roundabout

  36. Intersection

  37. Bridge

If you use this dataset in your research, please cite the following paper:

New in version 0.2.

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', transforms=None, checksum=False)[source]

Initialize a new FAIR1M dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

GID-15 (Gaofen Image Dataset)

class torchgeo.datasets.GID15(root='data', split='train', transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionDataset

GID-15 dataset.

The GID-15 dataset is a dataset for semantic segmentation.

Dataset features:

  • images taken by the Gaofen-2 (GF-2) satellite over 60 cities in China

  • masks representing 15 semantic categories

  • three spectral bands - RGB

  • 150 with 3 m per pixel resolution (6800x7200 px)

Dataset format:

  • images are three-channel pngs

  • masks are single-channel pngs

  • colormapped masks are 3 channel tifs

Dataset classes:

  1. background

  2. industrial_land

  3. urban_residential

  4. rural_residential

  5. traffic_land

  6. paddy_field

  7. irrigated_land

  8. dry_cropland

  9. garden_plot

  10. arbor_woodland

  11. shrub_land

  12. natural_grassland

  13. artificial_grassland

  14. river

  15. lake

  16. pond

If you use this dataset in your research, please cite the following paper:

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', split='train', transforms=None, download=False, checksum=False)[source]

Initialize a new GID-15 dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train”, “val”, or “test”

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises
  • AssertionError – if split argument is invalid

  • RuntimeError – if download=False and data is not found, or checksums don’t match

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns;

a matplotlib Figure with the rendered sample

New in version 0.2.

IDTReeS

class torchgeo.datasets.IDTReeS(root='data', split='train', task='task1', transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionDataset

IDTReeS dataset.

The IDTReeS dataset is a dataset for tree crown detection.

Dataset features:

  • RGB Image, Canopy Height Model (CHM), Hyperspectral Image (HSI), LiDAR Point Cloud

  • Remote sensing and field data generated by the National Ecological Observatory Network (NEON)

  • 0.1 - 1m resolution imagery

  • Task 1 - object detection (tree crown delination)

  • Task 2 - object classification (species classification)

  • Train set contains 85 images

  • Test set (task 1) contains 153 images

  • Test set (task 2) contains 353 images and tree crown polygons

Dataset format:

  • optical - three-channel RGB 200x200 geotiff

  • canopy height model - one-channel 20x20 geotiff

  • hyperspectral - 369-channel 20x20 geotiff

  • point cloud - Nx3 LAS file (.las), some files contain RGB colors per point

  • shapely files (.shp) containing polygons

  • csv file containing species labels and other metadata for each polygon

Dataset classes:

  1. ACPE

  2. ACRU

  3. ACSA3

  4. AMLA

  5. BETUL

  6. CAGL8

  7. CATO6

  8. FAGR

  9. GOLA

  10. LITU

  11. LYLU3

  12. MAGNO

  13. NYBI

  14. NYSY

  15. OXYDE

  16. PEPA37

  17. PIEL

  18. PIPA2

  19. PINUS

  20. PITA

  21. PRSE2

  22. QUAL

  23. QUCO2

  24. QUGE2

  25. QUHE2

  26. QULA2

  27. QULA3

  28. QUMO4

  29. QUNI

  30. QURU

  31. QUERC

  32. ROPS

  33. TSCA

If you use this dataset in your research, please cite the following paper:

New in version 0.2.

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', split='train', task='task1', transforms=None, download=False, checksum=False)[source]

Initialize a new IDTReeS dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train” or “test”

  • task (str) – ‘task1’ for detection, ‘task2’ for detection + classification (only relevant for split=’test’)

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises

ImportError – if laspy or pandas are are not installed

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, suptitle=None, hsi_indices=(0, 1, 2))[source]

Plot a sample from the dataset.

Parameters
  • sample (Dict[str, torch.Tensor]) – a sample returned by __getitem__()

  • show_titles (bool) – flag indicating whether to show titles above each panel

  • suptitle (Optional[str]) – optional string to use as a suptitle

  • hsi_indices (Tuple[int, int, int]) – tuple of indices to create HSI false color image

Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

plot_las(index, colormap=None)[source]

Plot a sample point cloud at the index.

Parameters
  • index (int) – index to plot

  • colormap (Optional[str]) – a valid matplotlib colormap

Returns

a open3d.visualizer.Visualizer object. Use

Visualizer.run() to display

Raises

ImportError – if open3d is not installed

Return type

Any

LandCover.ai (Land Cover from Aerial Imagery)

class torchgeo.datasets.LandCoverAI(root='data', split='train', transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionDataset

LandCover.ai dataset.

The LandCover.ai (Land Cover from Aerial Imagery) dataset is a dataset for automatic mapping of buildings, woodlands, water and roads from aerial images. This implementation is specifically for Version 1 of Landcover.ai.

Dataset features:

  • land cover from Poland, Central Europe

  • three spectral bands - RGB

  • 33 orthophotos with 25 cm per pixel resolution (~9000x9500 px)

  • 8 orthophotos with 50 cm per pixel resolution (~4200x4700 px)

  • total area of 216.27 km2

Dataset format:

  • rasters are three-channel GeoTiffs with EPSG:2180 spatial reference system

  • masks are single-channel GeoTiffs with EPSG:2180 spatial reference system

Dataset classes:

  1. building (1.85 km2)

  2. woodland (72.02 km2)

  3. water (13.15 km2)

  4. road (3.5 km2)

If you use this dataset in your research, please cite the following paper:

Note

This dataset requires the following additional library to be installed:

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', split='train', transforms=None, download=False, checksum=False)[source]

Initialize a new LandCover.ai dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train”, “val”, or “test”

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises
  • AssertionError – if split argument is invalid

  • RuntimeError – if download=False and data is not found, or checksums don’t match

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

New in version 0.2.

LEVIR-CD+ (LEVIR Change Detection +)

class torchgeo.datasets.LEVIRCDPlus(root='data', split='train', transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionDataset

LEVIR-CD+ dataset.

The LEVIR-CD+ dataset is a dataset for building change detection.

Dataset features:

  • image pairs of 20 different urban regions across Texas between 2002-2020

  • binary change masks representing building change

  • three spectral bands - RGB

  • 985 image pairs with 50 cm per pixel resolution (~1024x1024 px)

Dataset format:

  • images are three-channel pngs

  • masks are single-channel pngs where no change = 0, change = 255

Dataset classes:

  1. no change

  2. change

If you use this dataset in your research, please cite the following paper:

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', split='train', transforms=None, download=False, checksum=False)[source]

Initialize a new LEVIR-CD+ dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train” or “test”

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises
  • AssertionError – if split argument is invalid

  • RuntimeError – if download=False and data is not found, or checksums don’t match

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

New in version 0.2.

LoveDA (Land-cOVEr Domain Adaptive semantic segmentation)

class torchgeo.datasets.LoveDA(root='data', split='train', scene=['urban', 'rural'], transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionDataset

LoveDA dataset.

The LoveDA datataset is a semantic segmentation dataset.

Dataset features:

  • 2713 urban scene and 3274 rural scene HSR images, spatial resolution of 0.3m

  • image source is Google Earth platform

  • total of 166768 annotated objects from Nanjing, Changzhou and Wuhan cities

  • dataset comes with predefined train, validation, and test set

  • dataset differentiates between ‘rural’ and ‘urban’ images

Dataset format:

  • images are three-channel pngs with dimension 1024x1024

  • segmentation masks are single-channel pngs

Dataset classes:

  1. background

  2. building

  3. road

  4. water

  5. barren

  6. forest

  7. agriculture

No-data regions assigned with 0 and should be ignored.

If you use this dataset in your research, please cite the following paper:

New in version 0.2.

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

image and mask at that index with image of dimension 3x1024x1024 and mask of dimension 1024x1024

Return type

Dict[str, torch.Tensor]

__init__(root='data', split='train', scene=['urban', 'rural'], transforms=None, download=False, checksum=False)[source]

Initialize a new LoveDA dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train”, “val”, or “test”

  • scene (List[str]) – specify whether to load only ‘urban’, only ‘rural’ or both

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises
__len__()[source]

Return the number of datapoints in the dataset.

Returns

length of dataset

Return type

int

plot(sample, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

NASA Marine Debris

class torchgeo.datasets.NASAMarineDebris(root='data', transforms=None, download=False, api_key=None, checksum=False, verbose=False)

Bases: torchgeo.datasets.VisionDataset

NASA Marine Debris dataset.

The NASA Marine Debris dataset is a dataset for detection of floating marine debris in satellite imagery.

Dataset features:

  • 707 patches with 3 m per pixel resolution (256x256 px)

  • three spectral bands - RGB

  • 1 object class: marine_debris

  • images taken by Planet Labs PlanetScope satellites

  • imagery taken from 2016-2019 from coasts of Greece, Honduras, and Ghana

Dataset format:

  • images are three-channel geotiffs in uint8 format

  • labels are numpy files (.npy) containing bounding box (xyxy) coordinates

  • additional: images in jpg format and labels in geojson format

If you use this dataset in your research, please cite the following paper:

Note

This dataset requires the following additional library to be installed:

  • radiant-mlhub to download the imagery and labels from the Radiant Earth MLHub

New in version 0.2.

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and labels at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', transforms=None, download=False, api_key=None, checksum=False, verbose=False)[source]

Initialize a new NASA Marine Debris Dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • api_key (Optional[str]) – a RadiantEarth MLHub API key to use for downloading the dataset

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

  • verbose (bool) – if True, print messages when new tiles are loaded

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

OSCD (Onera Satellite Change Detection)

class torchgeo.datasets.OSCD(root='data', split='train', bands='all', transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionDataset

OSCD dataset.

The Onera Satellite Change Detection dataset addresses the issue of detecting changes between satellite images from different dates. Imagery comes from Sentinel-2 which contains varying resolutions per band.

Dataset format:

  • images are 13-channel tifs

  • masks are single-channel pngs where no change = 0, change = 255

Dataset classes:

  1. no change

  2. change

If you use this dataset in your research, please cite the following paper:

New in version 0.2.

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', split='train', bands='all', transforms=None, download=False, checksum=False)[source]

Initialize a new OSCD dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train” or “test”

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises
  • AssertionError – if split argument is invalid

  • RuntimeError – if download=False and data is not found, or checksums don’t match

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, suptitle=None, alpha=0.5)[source]

Plot a sample from the dataset.

Parameters
  • sample (Dict[str, torch.Tensor]) – a sample returned by __getitem__()

  • show_titles (bool) – flag indicating whether to show titles above each panel

  • suptitle (Optional[str]) – optional string to use as a suptitle

  • alpha (float) – opacity with which to render predictions on top of the imagery

Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

PatternNet

class torchgeo.datasets.PatternNet(root='data', transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionClassificationDataset

PatternNet dataset.

The PatternNet dataset is a dataset for remote sensing scene classification and image retrieval.

Dataset features:

  • 30,400 images with 6-50 cm per pixel resolution (256x256 px)

  • three spectral bands - RGB

  • 38 scene classes, 800 images per class

Dataset format:

  • images are three-channel jpgs

Dataset classes:

  1. airplane

  2. baseball_field

  3. basketball_court

  4. beach

  5. bridge

  6. cemetery

  7. chaparral

  8. christmas_tree_farm

  9. closed_road

  10. coastal_mansion

  11. crosswalk

  12. dense_residential

  13. ferry_terminal

  14. football_field

  15. forest

  16. freeway

  17. golf_course

  18. harbor

  19. intersection

  20. mobile_home_park

  21. nursing_home

  22. oil_gas_field

  23. oil_well

  24. overpass

  25. parking_lot

  26. parking_space

  27. railway

  28. river

  29. runway

  30. runway_marking

  31. shipping_yard

  32. solar_panel

  33. sparse_residential

  34. storage_tank

  35. swimming_pool

  36. tennis_court

  37. transformer_station

  38. wastewater_treatment_plant

If you use this dataset in your research, please cite the following paper:

__init__(root='data', transforms=None, download=False, checksum=False)[source]

Initialize a new PatternNet dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

New in version 0.2.

Potsdam

class torchgeo.datasets.Potsdam2D(root='data', split='train', transforms=None, checksum=False)

Bases: torchgeo.datasets.VisionDataset

Potsdam 2D Semantic Segmentation dataset.

The Potsdam dataset is a dataset for urban semantic segmentation used in the 2D Semantic Labeling Contest - Potsdam. This dataset uses the “4_Ortho_RGBIR.zip” and “5_Labels_all.zip” files to create the train/test sets used in the challenge. The dataset can be requested at the challenge homepage. Note, the server contains additional data for 3D Semantic Labeling which are currently not supported.

Dataset format:

  • images are 4-channel geotiffs

  • masks are 3-channel geotiffs with unique RGB values representing the class

Dataset classes:

  1. Clutter/background

  2. Impervious surfaces

  3. Building

  4. Low Vegetation

  5. Tree

  6. Car

If you use this dataset in your research, please cite the following paper:

New in version 0.2.

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', split='train', transforms=None, checksum=False)[source]

Initialize a new Potsdam dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train” or “test”

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, suptitle=None, alpha=0.5)[source]

Plot a sample from the dataset.

Parameters
  • sample (Dict[str, torch.Tensor]) – a sample returned by __getitem__()

  • show_titles (bool) – flag indicating whether to show titles above each panel

  • suptitle (Optional[str]) – optional string to use as a suptitle

  • alpha (float) – opacity with which to render predictions on top of the imagery

Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

RESISC45 (Remote Sensing Image Scene Classification)

class torchgeo.datasets.RESISC45(root='data', split='train', transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionClassificationDataset

RESISC45 dataset.

The RESISC45 dataset is a dataset for remote sensing image scene classification.

Dataset features:

  • 31,500 images with 0.2-30 m per pixel resolution (256x256 px)

  • three spectral bands - RGB

  • 45 scene classes, 700 images per class

  • images extracted from Google Earth from over 100 countries

  • images conditions with high variability (resolution, weather, illumination)

Dataset format:

  • images are three-channel jpgs

Dataset classes:

  1. airplane

  2. airport

  3. baseball_diamond

  4. basketball_court

  5. beach

  6. bridge

  7. chaparral

  8. church

  9. circular_farmland

  10. cloud

  11. commercial_area

  12. dense_residential

  13. desert

  14. forest

  15. freeway

  16. golf_course

  17. ground_track_field

  18. harbor

  19. industrial_area

  20. intersection

  21. island

  22. lake

  23. meadow

  24. medium_residential

  25. mobile_home_park

  26. mountain

  27. overpass

  28. palace

  29. parking_lot

  30. railway

  31. railway_station

  32. rectangular_farmland

  33. river

  34. roundabout

  35. runway

  36. sea_ice

  37. ship

  38. snowberg

  39. sparse_residential

  40. stadium

  41. storage_tank

  42. tennis_court

  43. terrace

  44. thermal_power_station

  45. wetland

This dataset uses the train/val/test splits defined in the “In-domain representation learning for remote sensing” paper:

If you use this dataset in your research, please cite the following paper:

__init__(root='data', split='train', transforms=None, download=False, checksum=False)[source]

Initialize a new RESISC45 dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train”, “val”, or “test”

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

New in version 0.2.

Seasonal Contrast

class torchgeo.datasets.SeasonalContrastS2(root='data', version='100k', bands=['B4', 'B3', 'B2'], transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionDataset

Sentinel 2 imagery from the Seasonal Contrast paper.

The Seasonal Contrast imagery dataset contains Sentinel 2 imagery patches sampled from different points in time around the 10k most populated cities on Earth.

Dataset features:

  • Two versions: 100K and 1M patches

  • 12 band Sentinel 2 imagery from 5 points in time at each location

If you use this dataset in your research, please cite the following paper:

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

sample with an “image” in 5xCxHxW format where the 5 indexes over the same

patch sampled from different points in time by the SeCo method

Return type

Dict[str, torch.Tensor]

__init__(root='data', version='100k', bands=['B4', 'B3', 'B2'], transforms=None, download=False, checksum=False)[source]

Initialize a new SeCo dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • version (str) – one of “100k” or “1m” for the version of the dataset to use

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises
  • AssertionError – if version argument is invalid

  • RuntimeError – if download=False and data is not found, or checksums don’t match

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Raises

ValueError – if the RGB bands are included in self.bands or the sample contains a “prediction” key

Return type

matplotlib.figure.Figure

New in version 0.2.

SEN12MS

class torchgeo.datasets.SEN12MS(root='data', split='train', bands=('VV', 'VH', 'B01', 'B02', 'B03', 'B04', 'B05', 'B06', 'B07', 'B08', 'B8A', 'B09', 'B10', 'B11', 'B12'), transforms=None, checksum=False)

Bases: torchgeo.datasets.VisionDataset

SEN12MS dataset.

The SEN12MS dataset contains 180,662 patch triplets of corresponding Sentinel-1 dual-pol SAR data, Sentinel-2 multi-spectral images, and MODIS-derived land cover maps. The patches are distributed across the land masses of the Earth and spread over all four meteorological seasons. This is reflected by the dataset structure. All patches are provided in the form of 16-bit GeoTiffs containing the following specific information:

  • Sentinel-1 SAR: 2 channels corresponding to sigma nought backscatter values in dB scale for VV and VH polarization.

  • Sentinel-2 Multi-Spectral: 13 channels corresponding to the 13 spectral bands (B1, B2, B3, B4, B5, B6, B7, B8, B8a, B9, B10, B11, B12).

  • MODIS Land Cover: 4 channels corresponding to IGBP, LCCS Land Cover, LCCS Land Use, and LCCS Surface Hydrology layers.

If you use this dataset in your research, please cite the following paper:

Note

This dataset can be automatically downloaded using the following bash script:

for season in 1158_spring 1868_summer 1970_fall 2017_winter
do
    for source in lc s1 s2
    do
        wget "ftp://m1474000:m1474000@dataserv.ub.tum.de/ROIs${season}_${source}.tar.gz"
        tar xvzf "ROIs${season}_${source}.tar.gz"
    done
done

for split in train test
do
    wget "https://raw.githubusercontent.com/schmitt-muc/SEN12MS/master/splits/${split}_list.txt"
done

or manually downloaded from https://dataserv.ub.tum.de/s/m1474000 and https://github.com/schmitt-muc/SEN12MS/tree/master/splits. This download will likely take several hours.

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', split='train', bands=('VV', 'VH', 'B01', 'B02', 'B03', 'B04', 'B05', 'B06', 'B07', 'B08', 'B8A', 'B09', 'B10', 'B11', 'B12'), transforms=None, checksum=False)[source]

Initialize a new SEN12MS dataset instance.

The bands argument allows for the subsetting of bands returned by the dataset. Integers in bands index into a stack of Sentinel 1 and Sentinel 2 imagery. Indices 0 and 1 correspond to the Sentinel 1 imagery where indices 2 through 14 correspond to the Sentinel 2 imagery.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train” or “test”

  • bands (Sequence[str]) – a sequence of band indices to use where the indices correspond to the array index of combined Sentinel 1 and Sentinel 2

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises
__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

New in version 0.2.

So2Sat

class torchgeo.datasets.So2Sat(root='data', split='train', transforms=None, checksum=False)

Bases: torchgeo.datasets.VisionDataset

So2Sat dataset.

The So2Sat dataset consists of corresponding synthetic aperture radar and multispectral optical image data acquired by the Sentinel-1 and Sentinel-2 remote sensing satellites, and a corresponding local climate zones (LCZ) label. The dataset is distributed over 42 cities across different continents and cultural regions of the world, and comes with a split into fully independent, non-overlapping training, validation, and test sets.

This implementation focuses on the 2nd version of the dataset as described in the author’s github repository https://github.com/zhu-xlab/So2Sat-LCZ42 and hosted at https://mediatum.ub.tum.de/1483140. This version is identical to the first version of the dataset but includes the test data. The splits are defined as follows:

  • Training: 42 cities around the world

  • Validation: western half of 10 other cities covering 10 cultural zones

  • Testing: eastern half of the 10 other cities

Dataset classes:

  1. Compact high rise

  2. Compact middle rise

  3. Compact low rise

  4. Open high rise

  5. Open mid rise

  6. Open low rise

  7. Lightweight low rise

  8. Large low rise

  9. Sparsely built

  10. Heavy industry

  11. Dense trees

  12. Scattered trees

  13. Bush, scrub

  14. Low plants

  15. Bare rock or paved

  16. Bare soil or sand

  17. Water

If you use this dataset in your research, please cite the following paper:

Note

This dataset can be automatically downloaded using the following bash script:

for split in training validation testing
do
    wget ftp://m1483140:m1483140@dataserv.ub.tum.de/$split.h5
done

or manually downloaded from https://dataserv.ub.tum.de/index.php/s/m1483140 This download will likely take several hours.

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', split='train', transforms=None, checksum=False)[source]

Initialize a new So2Sat dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train”, “validation”, or “test”

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises
__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

New in version 0.2.

SpaceNet

class torchgeo.datasets.SpaceNet(root, image, collections=[], transforms=None, download=False, api_key=None, checksum=False)

Bases: torchgeo.datasets.VisionDataset, abc.ABC

Abstract base class for the SpaceNet datasets.

The SpaceNet datasets are a set of datasets that all together contain >11M building footprints and ~20,000 km of road labels mapped over high-resolution satellite imagery obtained from a variety of sensors such as Worldview-2, Worldview-3 and Dove.

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root, image, collections=[], transforms=None, download=False, api_key=None, checksum=False)[source]

Initialize a new SpaceNet Dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • image (str) – image selection

  • collections (List[str]) – collection selection

  • transforms (Optional[Callable[[Dict[str, Any]], Dict[str, Any]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version.

  • download (bool) – if True, download dataset and store it in the root directory.

  • api_key (Optional[str]) – a RadiantEarth MLHub API key to use for downloading the dataset

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises

RuntimeError – if download=False but dataset is missing

__len__()[source]

Return the number of samples in the dataset.

Returns

length of the dataset

Return type

int

abstract property chip_size: Dict[str, Tuple[int, int]]

Mapping of images and their chip size.

abstract property collection_md5_dict: Dict[str, str]

Mapping of collection id and md5 checksum.

abstract property dataset_id: str

Dataset ID.

abstract property imagery: Dict[str, str]

Mapping of image identifier and filename.

abstract property label_glob: str

Label filename.

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

New in version 0.2.

class torchgeo.datasets.SpaceNet1(root, image='rgb', transforms=None, download=False, api_key=None, checksum=False)

Bases: torchgeo.datasets.SpaceNet

SpaceNet 1: Building Detection v1 Dataset.

SpaceNet 1 is a dataset of building footprints over the city of Rio de Janeiro.

Dataset features:

  • No. of images: 6940 (8 Band) + 6940 (RGB)

  • No. of polygons: 382,534 building labels

  • Area Coverage: 2544 sq km

  • GSD: 1 m (8 band), 50 cm (rgb)

  • Chip size: 101 x 110 (8 band), 406 x 438 (rgb)

Dataset format:

  • Imagery - Worldview-2 GeoTIFFs

    • 8Band.tif (Multispectral)

    • RGB.tif (Pansharpened RGB)

  • Labels - GeoJSON

    • labels.geojson

If you use this dataset in your research, please cite the following paper:

Note

This dataset requires the following additional library to be installed:

  • radiant-mlhub to download the imagery and labels from the Radiant Earth MLHub

__init__(root, image='rgb', transforms=None, download=False, api_key=None, checksum=False)[source]

Initialize a new SpaceNet 1 Dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • image (str) – image selection which must be “rgb” or “8band”

  • transforms (Optional[Callable[[Dict[str, Any]], Dict[str, Any]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version.

  • download (bool) – if True, download dataset and store it in the root directory.

  • api_key (Optional[str]) – a RadiantEarth MLHub API key to use for downloading the dataset

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises

RuntimeError – if download=False but dataset is missing

class torchgeo.datasets.SpaceNet2(root, image='PS-RGB', collections=[], transforms=None, download=False, api_key=None, checksum=False)

Bases: torchgeo.datasets.SpaceNet

SpaceNet 2: Building Detection v2 Dataset.

SpaceNet 2 is a dataset of building footprints over the cities of Las Vegas, Paris, Shanghai and Khartoum.

Collection features:

AOI

Area (km2)

# Images

# Buildings

Las Vegas

216

3850

151,367

Paris

1030

1148

23,816

Shanghai

1000

4582

92,015

Khartoum

765

1012

35,503

Imagery features:

PAN

MS

PS-MS

PS-RGB

GSD (m)

0.31

1.24

0.30

0.30

Chip size (px)

650 x 650

162 x 162

650 x 650

650 x 650

Dataset format:

  • Imagery - Worldview-3 GeoTIFFs

    • PAN.tif (Panchromatic)

    • MS.tif (Multispectral)

    • PS-MS (Pansharpened Multispectral)

    • PS-RGB (Pansharpened RGB)

  • Labels - GeoJSON

    • label.geojson

If you use this dataset in your research, please cite the following paper:

Note

This dataset requires the following additional library to be installed:

  • radiant-mlhub to download the imagery and labels from the Radiant Earth MLHub

__init__(root, image='PS-RGB', collections=[], transforms=None, download=False, api_key=None, checksum=False)[source]

Initialize a new SpaceNet 2 Dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • image (str) – image selection which must be in [“MS”, “PAN”, “PS-MS”, “PS-RGB”]

  • collections (List[str]) – collection selection which must be a subset of: [sn2_AOI_2_Vegas, sn2_AOI_3_Paris, sn2_AOI_4_Shanghai, sn2_AOI_5_Khartoum]

  • transforms (Optional[Callable[[Dict[str, Any]], Dict[str, Any]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory.

  • api_key (Optional[str]) – a RadiantEarth MLHub API key to use for downloading the dataset

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises

RuntimeError – if download=False but dataset is missing

class torchgeo.datasets.SpaceNet4(root, image='PS-RGBNIR', angles=[], transforms=None, download=False, api_key=None, checksum=False)

Bases: torchgeo.datasets.SpaceNet

SpaceNet 4: Off-Nadir Buildings Dataset.

SpaceNet 4 is a dataset of 27 WV-2 imagery captured at varying off-nadir angles and associated building footprints over the city of Atlanta. The off-nadir angle ranges from 7 degrees to 54 degrees.

Dataset features:

  • No. of chipped images: 28,728 (PAN/MS/PS-RGBNIR)

  • No. of label files: 1064

  • No. of building footprints: >120,000

  • Area Coverage: 665 sq km

  • Chip size: 225 x 225 (MS), 900 x 900 (PAN/PS-RGBNIR)

Dataset format:

  • Imagery - Worldview-2 GeoTIFFs

    • PAN.tif (Panchromatic)

    • MS.tif (Multispectral)

    • PS-RGBNIR (Pansharpened RGBNIR)

  • Labels - GeoJSON

    • labels.geojson

If you use this dataset in your research, please cite the following paper:

Note

This dataset requires the following additional library to be installed:

  • radiant-mlhub to download the imagery and labels from the Radiant Earth MLHub

__init__(root, image='PS-RGBNIR', angles=[], transforms=None, download=False, api_key=None, checksum=False)[source]

Initialize a new SpaceNet 4 Dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • image (str) – image selection which must be in [“MS”, “PAN”, “PS-RGBNIR”]

  • angles (List[str]) – angle selection which must be in [“nadir”, “off-nadir”, “very-off-nadir”]

  • transforms (Optional[Callable[[Dict[str, Any]], Dict[str, Any]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory.

  • api_key (Optional[str]) – a RadiantEarth MLHub API key to use for downloading the dataset

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises

RuntimeError – if download=False but dataset is missing

class torchgeo.datasets.SpaceNet5(root, image='PS-RGB', speed_mask=False, collections=[], transforms=None, download=False, api_key=None, checksum=False)

Bases: torchgeo.datasets.SpaceNet

SpaceNet 5: Automated Road Network Extraction and Route Travel Time Estimation.

SpaceNet 5 is a dataset of road networks over the cities of Moscow, Mumbai and San Juan (unavailable).

Collection features:

AOI

Area (km2)

# Images

# Road Network Labels (km)

Moscow

1353

1353

3066

Mumbai

1021

1016

1951

Imagery features:

PAN

MS

PS-MS

PS-RGB

GSD (m)

0.31

1.24

0.30

0.30

Chip size (px)

1300 x 1300

325 x 325

1300 x 1300

1300 x 1300

Dataset format:

  • Imagery - Worldview-3 GeoTIFFs

    • PAN.tif (Panchromatic)

    • MS.tif (Multispectral)

    • PS-MS (Pansharpened Multispectral)

    • PS-RGB (Pansharpened RGB)

  • Labels - GeoJSON

    • labels.geojson

If you use this dataset in your research, please use the following citation:

Note

This dataset requires the following additional library to be installed:

  • radiant-mlhub to download the imagery and labels from the Radiant Earth MLHub

New in version 0.2.

__init__(root, image='PS-RGB', speed_mask=False, collections=[], transforms=None, download=False, api_key=None, checksum=False)[source]

Initialize a new SpaceNet 5 Dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • image (str) – image selection which must be in [“MS”, “PAN”, “PS-MS”, “PS-RGB”]

  • speed_mask (Optional[bool]) – use multi-class speed mask (created by binning roads at 10 mph increments) as label if true, else use binary mask

  • collections (List[str]) – collection selection which must be a subset of: [sn5_AOI_7_Moscow, sn5_AOI_8_Mumbai]

  • transforms (Optional[Callable[[Dict[str, Any]], Dict[str, Any]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory.

  • api_key (Optional[str]) – a RadiantEarth MLHub API key to use for downloading the dataset

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises

RuntimeError – if download=False but dataset is missing

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

New in version 0.2.

class torchgeo.datasets.SpaceNet7(root, split='train', transforms=None, download=False, api_key=None, checksum=False)

Bases: torchgeo.datasets.SpaceNet

SpaceNet 7: Multi-Temporal Urban Development Challenge.

SpaceNet 7 is a dataset which consist of medium resolution (4.0m) satellite imagery mosaics acquired from Planet Labs’ Dove constellation between 2017 and 2020. It includes ≈ 24 images (one per month) covering > 100 unique geographies, and comprises > 40,000 km2 of imagery and exhaustive polygon labels of building footprints therein, totaling over 11M individual annotations.

Dataset features:

  • No. of train samples: 1423

  • No. of test samples: 466

  • No. of building footprints: 11,080,000

  • Area Coverage: 41,000 sq km

  • Chip size: 1023 x 1023

  • GSD: ~4m

Dataset format:

  • Imagery - Planet Dove GeoTIFF

    • mosaic.tif

  • Labels - GeoJSON

    • labels.geojson

If you use this dataset in your research, please cite the following paper:

Note

This dataset requires the following additional library to be installed:

  • radiant-mlhub to download the imagery and labels from the Radiant Earth MLHub

New in version 0.2.

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data at that index

Return type

Dict[str, torch.Tensor]

__init__(root, split='train', transforms=None, download=False, api_key=None, checksum=False)[source]

Initialize a new SpaceNet 7 Dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – split selection which must be in [“train”, “test”]

  • transforms (Optional[Callable[[Dict[str, Any]], Dict[str, Any]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory.

  • api_key (Optional[str]) – a RadiantEarth MLHub API key to use for downloading the dataset

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises

RuntimeError – if download=False but dataset is missing

Tropical Cyclone Wind Estimation Competition

class torchgeo.datasets.TropicalCycloneWindEstimation(root='data', split='train', transforms=None, download=False, api_key=None, checksum=False)

Bases: torchgeo.datasets.VisionDataset

Tropical Cyclone Wind Estimation Competition dataset.

A collection of tropical storms in the Atlantic and East Pacific Oceans from 2000 to 2019 with corresponding maximum sustained surface wind speed. This dataset is split into training and test categories for the purpose of a competition.

See https://www.drivendata.org/competitions/72/predict-wind-speeds/ for more information about the competition.

If you use this dataset in your research, please cite the following paper:

Note

This dataset requires the following additional library to be installed:

  • radiant-mlhub to download the imagery and labels from the Radiant Earth MLHub

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data, labels, field ids, and metadata at that index

Return type

Dict[str, Any]

__init__(root='data', split='train', transforms=None, download=False, api_key=None, checksum=False)[source]

Initialize a new Tropical Cyclone Wind Estimation Competition Dataset.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train” or “test”

  • transforms (Optional[Callable[[Dict[str, Any]], Dict[str, Any]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • api_key (Optional[str]) – a RadiantEarth MLHub API key to use for downloading the dataset

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises
__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns;

a matplotlib Figure with the rendered sample

New in version 0.2.

UC Merced

class torchgeo.datasets.UCMerced(root='data', split='train', transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionClassificationDataset

UC Merced dataset.

The UC Merced dataset is a land use classification dataset of 2.1k 256x256 1ft resolution RGB images of urban locations around the U.S. extracted from the USGS National Map Urban Area Imagery collection with 21 land use classes (100 images per class).

Dataset features:

  • land use class labels from around the U.S.

  • three spectral bands - RGB

  • 21 classes

Dataset classes:

  • agricultural

  • airplane

  • baseballdiamond

  • beach

  • buildings

  • chaparral

  • denseresidential

  • forest

  • freeway

  • golfcourse

  • harbor

  • intersection

  • mediumresidential

  • mobilehomepark

  • overpass

  • parkinglot

  • river

  • runway

  • sparseresidential

  • storagetanks

  • tenniscourt

This dataset uses the train/val/test splits defined in the “In-domain representation learning for remote sensing” paper:

If you use this dataset in your research, please cite the following paper:

__init__(root='data', split='train', transforms=None, download=False, checksum=False)[source]

Initialize a new UC Merced dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train”, “val”, or “test”

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises

RuntimeError – if download=False and data is not found, or checksums don’t match

plot(sample, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

New in version 0.2.

Vaihingen

class torchgeo.datasets.Vaihingen2D(root='data', split='train', transforms=None, checksum=False)

Bases: torchgeo.datasets.VisionDataset

Vaihingen 2D Semantic Segmentation dataset.

The Vaihingen dataset is a dataset for urban semantic segmentation used in the 2D Semantic Labeling Contest - Vaihingen. This dataset uses the “ISPRS_semantic_labeling_Vaihingen.zip” and “ISPRS_semantic_labeling_Vaihingen_ground_truth_COMPLETE.zip” files to create the train/test sets used in the challenge. The dataset can be requested at the challenge homepage. Note, the server contains additional data for 3D Semantic Labeling which are currently not supported.

Dataset format:

  • images are 3-channel RGB geotiffs

  • masks are 3-channel geotiffs with unique RGB values representing the class

Dataset classes:

  1. Clutter/background

  2. Impervious surfaces

  3. Building

  4. Low Vegetation

  5. Tree

  6. Car

If you use this dataset in your research, please cite the following paper:

New in version 0.2.

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', split='train', transforms=None, checksum=False)[source]

Initialize a new Vaihingen2D dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train” or “test”

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, suptitle=None, alpha=0.5)[source]

Plot a sample from the dataset.

Parameters
  • sample (Dict[str, torch.Tensor]) – a sample returned by __getitem__()

  • show_titles (bool) – flag indicating whether to show titles above each panel

  • suptitle (Optional[str]) – optional string to use as a suptitle

  • alpha (float) – opacity with which to render predictions on top of the imagery

Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

NWPU VHR-10

class torchgeo.datasets.VHR10(root='data', split='positive', transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionDataset

NWPU VHR-10 dataset.

Northwestern Polytechnical University (NWPU) very-high-resolution ten-class (VHR-10) remote sensing image dataset.

Consists of 800 VHR optical remote sensing images, where 715 color images were acquired from Google Earth with the spatial resolution ranging from 0.5 to 2 m, and 85 pansharpened color infrared (CIR) images were acquired from Vaihingen data with a spatial resolution of 0.08 m.

The data set is divided into two sets:

  • Positive image set (650 images) which contains at least one target in an image

  • Negative image set (150 images) does not contain any targets

The positive image set consists of objects from ten classes:

  1. Airplanes (757)

  2. Ships (302)

  3. Storage tanks (655)

  4. Baseball diamonds (390)

  5. Tennis courts (524)

  6. Basketball courts (159)

  7. Ground track fields (163)

  8. Harbors (224)

  9. Bridges (124)

  10. Vehicles (477)

Includes object detection bounding boxes from original paper and instance segmentation masks from follow-up publications. If you use this dataset in your research, please cite the following papers:

Note

This dataset requires the following additional libraries to be installed:

  • pycocotools to load the annotations.json file for the “positive” image set

  • rarfile to extract the dataset, which is stored in a RAR file

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, Any]

__init__(root='data', split='positive', transforms=None, download=False, checksum=False)[source]

Initialize a new VHR-10 dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “postive” or “negative”

  • transforms (Optional[Callable[[Dict[str, Any]], Dict[str, Any]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises
  • AssertionError – if split argument is invalid

  • RuntimeError – if download=False and data is not found, or checksums don’t match

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

xView2

class torchgeo.datasets.XView2(root='data', split='train', transforms=None, checksum=False)

Bases: torchgeo.datasets.VisionDataset

xView2 dataset.

The xView2 dataset is a dataset for building disaster change detection. This dataset object uses the “Challenge training set (~7.8 GB)” and “Challenge test set (~2.6 GB)” data from the xView2 website as the train and test splits. Note, the xView2 website contains other data under the xView2 umbrella that are _not_ included here. E.g. the “Tier3 training data”, the “Challenge holdout set”, and the “full data”.

Dataset format:

  • images are three-channel pngs

  • masks are single-channel pngs where the pixel values represent the class

Dataset classes:

  1. background

  2. no damage

  3. minor damage

  4. major damage

  5. destroyed

If you use this dataset in your research, please cite the following paper:

New in version 0.2.

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root='data', split='train', transforms=None, checksum=False)[source]

Initialize a new xView2 dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • split (str) – one of “train” or “test”

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, show_titles=True, suptitle=None, alpha=0.5)[source]

Plot a sample from the dataset.

Parameters
  • sample (Dict[str, torch.Tensor]) – a sample returned by __getitem__()

  • show_titles (bool) – flag indicating whether to show titles above each panel

  • suptitle (Optional[str]) – optional string to use as a suptitle

  • alpha (float) – opacity with which to render predictions on top of the imagery

Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

ZueriCrop

class torchgeo.datasets.ZueriCrop(root='data', bands=('NIR', 'B03', 'B02', 'B04', 'B05', 'B06', 'B07', 'B11', 'B12'), transforms=None, download=False, checksum=False)

Bases: torchgeo.datasets.VisionDataset

ZueriCrop dataset.

The ZueriCrop dataset is a dataset for time-series instance segmentation of crops.

Dataset features:

  • Sentinel-2 multispectral imagery

  • instance masks of 48 crop categories

  • nine multispectral bands

  • 116k images with 10 m per pixel resolution (24x24 px)

  • ~28k time-series containing 142 images each

Dataset format:

  • single hdf5 dataset containing images, semantic masks, and instance masks

  • data is parsed into images and instance masks, boxes, and labels

  • one mask per time-series

Dataset classes:

If you use this dataset in your research, please cite the following paper:

Note

This dataset requires the following additional library to be installed:

  • h5py to load the dataset

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

sample containing image, mask, bounding boxes, and target label

Return type

Dict[str, torch.Tensor]

__init__(root='data', bands=('NIR', 'B03', 'B02', 'B04', 'B05', 'B06', 'B07', 'B11', 'B12'), transforms=None, download=False, checksum=False)[source]

Initialize a new ZueriCrop dataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • bands (Sequence[str]) – the subset of bands to load

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • download (bool) – if True, download dataset and store it in the root directory

  • checksum (bool) – if True, check the MD5 of the downloaded files (may be slow)

Raises

RuntimeError – if download=False and data is not found, or checksums don’t match

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

plot(sample, time_step=0, show_titles=True, suptitle=None)[source]

Plot a sample from the dataset.

Parameters
  • sample (Dict[str, torch.Tensor]) – a sample returned by __getitem__()

  • time_step (int) – time step at which to access image, beginning with 0

  • show_titles (bool) – flag indicating whether to show titles above each panel

  • suptitle (Optional[str]) – optional suptitle to use for figure

Returns

a matplotlib Figure with the rendered sample

Return type

matplotlib.figure.Figure

New in version 0.2.

Base Classes

If you want to write your own custom dataset, you can extend one of these abstract base classes.

GeoDataset

class torchgeo.datasets.GeoDataset(transforms=None)

Bases: torch.utils.data.Dataset[Dict[str, Any]], abc.ABC

Abstract base class for datasets containing geospatial information.

Geospatial information includes things like:

GeoDataset is a special class of datasets. Unlike VisionDataset, the presence of geospatial information allows two or more datasets to be combined based on latitude/longitude. This allows users to do things like:

  • Combine image and target labels and sample from both simultaneously (e.g. Landsat and CDL)

  • Combine datasets for multiple image sources for multimodal learning or data fusion (e.g. Landsat and Sentinel)

These combinations require that all queries are present in both datasets, and can be combined using an IntersectionDataset:

dataset = landsat & cdl

Users may also want to:

  • Combine datasets for multiple image sources and treat them as equivalent (e.g. Landsat 7 and Landsat 8)

  • Combine datasets for disparate geospatial locations (e.g. Chesapeake NY and PA)

These combinations require that all queries are present in at least one dataset, and can be combined using a UnionDataset:

dataset = landsat7 | landsat8
__and__(other)[source]

Take the intersection of two GeoDataset.

Parameters

other (torchgeo.datasets.GeoDataset) – another dataset

Returns

a single dataset

Raises

ValueError – if other is not a GeoDataset

Return type

torchgeo.datasets.IntersectionDataset

New in version 0.2.

abstract __getitem__(query)[source]

Retrieve image/mask and metadata indexed by query.

Parameters

query (torchgeo.datasets.BoundingBox) – (minx, maxx, miny, maxy, mint, maxt) coordinates to index

Returns

sample of image/mask and metadata at that index

Raises

IndexError – if query is not found in the index

Return type

Dict[str, Any]

__getstate__()[source]

Define how instances are pickled.

Returns

the state necessary to unpickle the instance

Return type

Tuple[Dict[Any, Any], List[Tuple[int, Tuple[float, float, float, float, float, float], str]]]

__init__(transforms=None)[source]

Initialize a new Dataset instance.

Parameters

transforms (Optional[Callable[[Dict[str, Any]], Dict[str, Any]]]) – a function/transform that takes an input sample and returns a transformed version

__len__()[source]

Return the number of files in the dataset.

Returns

length of the dataset

Return type

int

__or__(other)[source]

Take the union of two GeoDatasets.

Parameters

other (torchgeo.datasets.GeoDataset) – another dataset

Returns

a single dataset

Raises

ValueError – if other is not a GeoDataset

Return type

torchgeo.datasets.UnionDataset

New in version 0.2.

__setstate__(state)[source]

Define how to unpickle an instance.

Parameters

state (Tuple[Dict[Any, Any], List[Tuple[int, Tuple[float, float, float, float, float, float], str]]]) – the state of the instance when it was pickled

__str__()[source]

Return the informal string representation of the object.

Returns

informal string representation

Return type

str

property bounds: torchgeo.datasets.BoundingBox

Bounds of the index.

Returns

(minx, maxx, miny, maxy, mint, maxt) of the dataset

property crs: rasterio.crs.CRS

coordinate reference system (CRS) for the dataset.

Returns

the coordinate reference system (CRS)

New in version 0.2.

RasterDataset

class torchgeo.datasets.RasterDataset(root, crs=None, res=None, transforms=None, cache=True)

Bases: torchgeo.datasets.GeoDataset

Abstract base class for GeoDataset stored as raster files.

__getitem__(query)[source]

Retrieve image/mask and metadata indexed by query.

Parameters

query (torchgeo.datasets.BoundingBox) – (minx, maxx, miny, maxy, mint, maxt) coordinates to index

Returns

sample of image/mask and metadata at that index

Raises

IndexError – if query is not found in the index

Return type

Dict[str, Any]

__init__(root, crs=None, res=None, transforms=None, cache=True)[source]

Initialize a new Dataset instance.

Parameters
Raises

FileNotFoundError – if no files are found in root

plot(data)[source]

Plot a data sample.

Parameters

data (torch.Tensor) – the data to plot

Raises

AssertionError – if is_image is True and data has a different number of channels than expected

VectorDataset

class torchgeo.datasets.VectorDataset(root='data', crs=None, res=0.0001, transforms=None)

Bases: torchgeo.datasets.GeoDataset

Abstract base class for GeoDataset stored as vector files.

__getitem__(query)[source]

Retrieve image/mask and metadata indexed by query.

Parameters

query (torchgeo.datasets.BoundingBox) – (minx, maxx, miny, maxy, mint, maxt) coordinates to index

Returns

sample of image/mask and metadata at that index

Raises

IndexError – if query is not found in the index

Return type

Dict[str, Any]

__init__(root='data', crs=None, res=0.0001, transforms=None)[source]

Initialize a new Dataset instance.

Parameters
Raises

FileNotFoundError – if no files are found in root

plot(data)[source]

Plot a data sample.

Parameters

data (torch.Tensor) – the data to plot

VisionDataset

class torchgeo.datasets.VisionDataset

Bases: torch.utils.data.Dataset[Dict[str, Any]], abc.ABC

Abstract base class for datasets lacking geospatial information.

This base class is designed for datasets with pre-defined image chips.

abstract __getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and labels at that index

Raises

IndexError – if index is out of range of the dataset

Return type

Dict[str, Any]

abstract __len__()[source]

Return the length of the dataset.

Returns

length of the dataset

Return type

int

__str__()[source]

Return the informal string representation of the object.

Returns

informal string representation

Return type

str

VisionClassificationDataset

class torchgeo.datasets.VisionClassificationDataset(root, transforms=None, loader=<function default_loader>, is_valid_file=None)

Bases: torchgeo.datasets.VisionDataset, torchvision.datasets.ImageFolder

Abstract base class for classification datasets lacking geospatial information.

This base class is designed for datasets with pre-defined image chips which are separated into separate folders per class.

__getitem__(index)[source]

Return an index within the dataset.

Parameters

index (int) – index to return

Returns

data and label at that index

Return type

Dict[str, torch.Tensor]

__init__(root, transforms=None, loader=<function default_loader>, is_valid_file=None)[source]

Initialize a new VisionClassificationDataset instance.

Parameters
  • root (str) – root directory where dataset can be found

  • transforms (Optional[Callable[[Dict[str, torch.Tensor]], Dict[str, torch.Tensor]]]) – a function/transform that takes input sample and its target as entry and returns a transformed version

  • loader (Optional[Callable[[str], Any]]) – a callable function which takes as input a path to an image and returns a PIL Image or numpy array

  • is_valid_file (Optional[Callable[[str], bool]]) – A function that takes the path of an Image file and checks if the file is a valid file

__len__()[source]

Return the number of data points in the dataset.

Returns

length of the dataset

Return type

int

IntersectionDataset

class torchgeo.datasets.IntersectionDataset(dataset1, dataset2, collate_fn=<function concat_samples>)

Bases: torchgeo.datasets.GeoDataset

Dataset representing the intersection of two GeoDatasets.

This allows users to do things like:

  • Combine image and target labels and sample from both simultaneously (e.g. Landsat and CDL)

  • Combine datasets for multiple image sources for multimodal learning or data fusion (e.g. Landsat and Sentinel)

These combinations require that all queries are present in both datasets, and can be combined using an IntersectionDataset:

dataset = landsat & cdl

New in version 0.2.

__getitem__(query)[source]

Retrieve image and metadata indexed by query.

Parameters

query (torchgeo.datasets.BoundingBox) – (minx, maxx, miny, maxy, mint, maxt) coordinates to index

Returns

sample of data/labels and metadata at that index

Raises

IndexError – if query is not within bounds of the index

Return type

Dict[str, Any]

__init__(dataset1, dataset2, collate_fn=<function concat_samples>)[source]

Initialize a new Dataset instance.

Parameters
Raises

ValueError – if either dataset is not a GeoDataset

__str__()[source]

Return the informal string representation of the object.

Returns

informal string representation

Return type

str

UnionDataset

class torchgeo.datasets.UnionDataset(dataset1, dataset2, collate_fn=<function merge_samples>)

Bases: torchgeo.datasets.GeoDataset

Dataset representing the union of two GeoDatasets.

This allows users to do things like:

  • Combine datasets for multiple image sources and treat them as equivalent (e.g. Landsat 7 and Landsat 8)

  • Combine datasets for disparate geospatial locations (e.g. Chesapeake NY and PA)

These combinations require that all queries are present in at least one dataset, and can be combined using a UnionDataset:

dataset = landsat7 | landsat8

New in version 0.2.

__getitem__(query)[source]

Retrieve image and metadata indexed by query.

Parameters

query (torchgeo.datasets.BoundingBox) – (minx, maxx, miny, maxy, mint, maxt) coordinates to index

Returns

sample of data/labels and metadata at that index

Raises

IndexError – if query is not within bounds of the index

Return type

Dict[str, Any]

__init__(dataset1, dataset2, collate_fn=<function merge_samples>)[source]

Initialize a new Dataset instance.

Parameters
Raises

ValueError – if either dataset is not a GeoDataset

__str__()[source]

Return the informal string representation of the object.

Returns

informal string representation

Return type

str

Utilities

class torchgeo.datasets.BoundingBox(minx, maxx, miny, maxy, mint, maxt)

Bases: object

Data class for indexing spatiotemporal data.

__and__(other)[source]

The intersection operator.

Parameters

other (torchgeo.datasets.BoundingBox) – another bounding box

Returns

the intersection of self and other

Raises

ValueError – if self and other do not intersect

Return type

torchgeo.datasets.BoundingBox

New in version 0.2.

__contains__(other)[source]

Whether or not other is within the bounds of this bounding box.

Parameters

other (torchgeo.datasets.BoundingBox) – another bounding box

Returns

True if other is within this bounding box, else False

Return type

bool

New in version 0.2.

__delattr__(name)

Implement delattr(self, name).

__eq__(other)

Return self==value.

__getitem__(key)[source]

Index the (minx, maxx, miny, maxy, mint, maxt) tuple.

Parameters

key (Union[int, slice]) – integer or slice object

Returns

the value(s) at that index

Raises

IndexError – if key is out of bounds

Return type

Union[float, List[float]]

__hash__()

Return hash(self).

__init__(minx, maxx, miny, maxy, mint, maxt)
__iter__()[source]

Container iterator.

Returns

iterator object that iterates over all objects in the container

Return type

Iterator[float]

__or__(other)[source]

The union operator.

Parameters

other (torchgeo.datasets.BoundingBox) – another bounding box

Returns

the minimum bounding box that contains both self and other

Return type

torchgeo.datasets.BoundingBox

New in version 0.2.

__post_init__()[source]

Validate the arguments passed to __init__().

Raises

ValueError – if bounding box is invalid (minx > maxx, miny > maxy, or mint > maxt)

New in version 0.2.

__repr__()

Return repr(self).

__setattr__(name, value)

Implement setattr(self, name, value).

__weakref__

list of weak references to the object (if defined)

intersects(other)[source]

Whether or not two bounding boxes intersect.

Parameters

other (torchgeo.datasets.BoundingBox) – another bounding box

Returns

True if bounding boxes intersect, else False

Return type

bool

Collation Functions

torchgeo.datasets.stack_samples(samples)

Stack a list of samples along a new axis.

Useful for forming a mini-batch of samples to pass to torch.utils.data.DataLoader.

Parameters

samples (Iterable[Dict[Any, Any]]) – list of samples

Returns

a single sample

Return type

Dict[Any, Any]

New in version 0.2.

torchgeo.datasets.concat_samples(samples)

Concatenate a list of samples along an existing axis.

Useful for joining samples in a torchgeo.datasets.IntersectionDataset.

Parameters

samples (Iterable[Dict[Any, Any]]) – list of samples

Returns

a single sample

Return type

Dict[Any, Any]

New in version 0.2.

torchgeo.datasets.merge_samples(samples)

Merge a list of samples.

Useful for joining samples in a torchgeo.datasets.UnionDataset.

Parameters

samples (Iterable[Dict[Any, Any]]) – list of samples

Returns

a single sample

Return type

Dict[Any, Any]

New in version 0.2.

torchgeo.datasets.unbind_samples(sample)

Reverse of stack_samples().

Useful for turning a mini-batch of samples into a list of samples. These individual samples can then be plotted using a dataset’s plot method.

Parameters

sample (Dict[Any, Sequence[Any]]) – a mini-batch of samples

Returns

list of samples

Return type

List[Dict[Any, Any]]

New in version 0.2.

Read the Docs v: v0.2.1
Versions
latest
stable
v0.2.1
v0.2.0
v0.1.1
v0.1.0
Downloads
On Read the Docs
Project Home
Builds

Free document hosting provided by Read the Docs.

Docs

Access comprehensive developer documentation for PyTorch

View Docs

Tutorials

Get in-depth tutorials for beginners and advanced developers

View Tutorials

Resources

Find development resources and get your questions answered

View Resources