data

data utils

---

source

GeoImageBlock

 GeoImageBlock (chnls_first=True)

Returns a TransformBlock that creates a GeoTensorImage object from an input image tensor.

Parameters

• chnls_first (bool, optional): A boolean indicating whether the input tensor has channels as the first dimension. Default is True.

Returns

• fastai.transforms.TransformBlock : A TransformBlock object that applies GeoTensorImage.create on an input tensor or filename.

Create a fastai datablock with geotiffs

from torchgeo.datamodules import EuroSATDataModule
from torchgeo.datasets import EuroSAT100
import fastai.vision.all as fv

# get sample geotiff dataset
sat_path = fv.untar_data(EuroSAT100.url)

# create a fastai datablock for geotiff classification 
dblock = fv.DataBlock(blocks=(GeoImageBlock(), fv.CategoryBlock()),
                      get_items=fv.get_image_files,
                      splitter=fv.RandomSplitter(valid_pct=0.1, seed=42),
                      get_y=fv.parent_label,
                      item_tfms=fv.Resize(64),
                      batch_tfms=[fv.Normalize.from_stats(EuroSATDataModule.mean, EuroSATDataModule.std)],
                     )

dblock.summary(sat_path, show_batch=True)

Setting-up type transforms pipelines
Collecting items from /home/butch2/.fastai/data/EuroSAT100
Found 100 items
2 datasets of sizes 90,10
Setting up Pipeline: partial
Setting up Pipeline: parent_label -> Categorize -- {'vocab': None, 'sort': True, 'add_na': False}

Building one sample
  Pipeline: partial
    starting from
      /home/butch2/.fastai/data/EuroSAT100/images/remote_sensing/otherDatasets/sentinel_2/tif/Industrial/Industrial_1906.tif
    applying partial gives
      GeoTensorImage of size 13x64x64
  Pipeline: parent_label -> Categorize -- {'vocab': None, 'sort': True, 'add_na': False}
    starting from
      /home/butch2/.fastai/data/EuroSAT100/images/remote_sensing/otherDatasets/sentinel_2/tif/Industrial/Industrial_1906.tif
    applying parent_label gives
      Industrial
    applying Categorize -- {'vocab': None, 'sort': True, 'add_na': False} gives
      TensorCategory(4)

Final sample: (GeoTensorImage: torch.Size([13, 64, 64]), TensorCategory(4))


Collecting items from /home/butch2/.fastai/data/EuroSAT100
Found 100 items
2 datasets of sizes 90,10
Setting up Pipeline: partial
Setting up Pipeline: parent_label -> Categorize -- {'vocab': None, 'sort': True, 'add_na': False}
Setting up after_item: Pipeline: Resize -- {'size': (64, 64), 'method': 'crop', 'pad_mode': 'reflection', 'resamples': (<Resampling.BILINEAR: 2>, <Resampling.NEAREST: 0>), 'p': 1.0} -> ToTensor
Setting up before_batch: Pipeline: 
Setting up after_batch: Pipeline: Normalize -- {'mean': tensor([[[[1354.4055]],

         [[1118.2440]],

         [[1042.9298]],

         [[ 947.6262]],

         [[1199.4729]],

         [[1999.7909]],

         [[2369.2229]],

         [[2296.8262]],

         [[ 732.0834]],

         [[  12.1133]],

         [[1819.0103]],

         [[1118.9240]],

         [[2594.1409]]]]), 'std': tensor([[[[ 245.7176]],

         [[ 333.0078]],

         [[ 395.0925]],

         [[ 593.7505]],

         [[ 566.4170]],

         [[ 861.1840]],

         [[1086.6313]],

         [[1117.9817]],

         [[ 404.9198]],

         [[   4.7758]],

         [[1002.5877]],

         [[ 761.3032]],

         [[1231.5858]]]]), 'axes': (0, 2, 3)}

Building one batch
Applying item_tfms to the first sample:
  Pipeline: Resize -- {'size': (64, 64), 'method': 'crop', 'pad_mode': 'reflection', 'resamples': (<Resampling.BILINEAR: 2>, <Resampling.NEAREST: 0>), 'p': 1.0} -> ToTensor
    starting from
      (GeoTensorImage of size 13x64x64, TensorCategory(4))
    applying Resize -- {'size': (64, 64), 'method': 'crop', 'pad_mode': 'reflection', 'resamples': (<Resampling.BILINEAR: 2>, <Resampling.NEAREST: 0>), 'p': 1.0} gives
      (GeoTensorImage of size 13x64x64, TensorCategory(4))
    applying ToTensor gives
      (GeoTensorImage of size 13x64x64, TensorCategory(4))

Adding the next 3 samples

No before_batch transform to apply

Collating items in a batch

Applying batch_tfms to the batch built
  Pipeline: Normalize -- {'mean': tensor([[[[1354.4055]],

         [[1118.2440]],

         [[1042.9298]],

         [[ 947.6262]],

         [[1199.4729]],

         [[1999.7909]],

         [[2369.2229]],

         [[2296.8262]],

         [[ 732.0834]],

         [[  12.1133]],

         [[1819.0103]],

         [[1118.9240]],

         [[2594.1409]]]]), 'std': tensor([[[[ 245.7176]],

         [[ 333.0078]],

         [[ 395.0925]],

         [[ 593.7505]],

         [[ 566.4170]],

         [[ 861.1840]],

         [[1086.6313]],

         [[1117.9817]],

         [[ 404.9198]],

         [[   4.7758]],

         [[1002.5877]],

         [[ 761.3032]],

         [[1231.5858]]]]), 'axes': (0, 2, 3)}
    starting from
      (GeoTensorImage of size 4x13x64x64, TensorCategory([4, 3, 8, 8]))
    applying Normalize -- {'mean': tensor([[[[1354.4055]],

         [[1118.2440]],

         [[1042.9298]],

         [[ 947.6262]],

         [[1199.4729]],

         [[1999.7909]],

         [[2369.2229]],

         [[2296.8262]],

         [[ 732.0834]],

         [[  12.1133]],

         [[1819.0103]],

         [[1118.9240]],

         [[2594.1409]]]]), 'std': tensor([[[[ 245.7176]],

         [[ 333.0078]],

         [[ 395.0925]],

         [[ 593.7505]],

         [[ 566.4170]],

         [[ 861.1840]],

         [[1086.6313]],

         [[1117.9817]],

         [[ 404.9198]],

         [[   4.7758]],

         [[1002.5877]],

         [[ 761.3032]],

         [[1231.5858]]]]), 'axes': (0, 2, 3)} gives
      (GeoTensorImage of size 4x13x64x64, TensorCategory([4, 3, 8, 8]))

dls = dblock.dataloaders(sat_path, bs=64)

dls.show_batch()