Train a Land Cover Classification Model¶

Install packages¶

To use the new functionality, ensure the required packages are installed.

In [ ]:

# %pip install geoai-py

# %pip install geoai-py

Import libraries¶

In [ ]:

import geoai

import geoai

Download sample data¶

In [ ]:

train_raster_url = "https://huggingface.co/datasets/giswqs/geospatial/resolve/main/m_3807511_ne_18_060_20181104.tif"
train_landcover_url = "https://huggingface.co/datasets/giswqs/geospatial/resolve/main/m_3807511_ne_18_060_20181104_landcover.tif"
test_raster_url = "https://huggingface.co/datasets/giswqs/geospatial/resolve/main/m_3807511_se_18_060_20181104.tif"

train_raster_url = "https://huggingface.co/datasets/giswqs/geospatial/resolve/main/m_3807511_ne_18_060_20181104.tif" train_landcover_url = "https://huggingface.co/datasets/giswqs/geospatial/resolve/main/m_3807511_ne_18_060_20181104_landcover.tif" test_raster_url = "https://huggingface.co/datasets/giswqs/geospatial/resolve/main/m_3807511_se_18_060_20181104.tif"

In [ ]:

train_raster_path = geoai.download_file(train_raster_url)
train_landcover_path = geoai.download_file(train_landcover_url)
test_raster_path = geoai.download_file(test_raster_url)

train_raster_path = geoai.download_file(train_raster_url) train_landcover_path = geoai.download_file(train_landcover_url) test_raster_path = geoai.download_file(test_raster_url)

Visualize sample data¶

In [ ]:

geoai.view_raster(train_landcover_url, basemap=train_raster_url)

geoai.view_raster(train_landcover_url, basemap=train_raster_url)

In [ ]:

geoai.view_raster(test_raster_url)

geoai.view_raster(test_raster_url)

Create training data¶

We will use the NAIP dataset for land cover classification. The classification scheme is adopted from the Chesapeake Land Cover project.

Important Note for Land Cover Classification:

• Your label images should contain integer class values (0, 1, 2, ..., 13 for 13 classes)
• Do NOT use binary masks - the training code now properly handles multi-class labels
• Class 0 is typically background, classes 1-12 are your land cover types

In [ ]:

out_folder = "landcover"
tiles = geoai.export_geotiff_tiles(
    in_raster=train_raster_path,
    out_folder=out_folder,
    in_class_data=train_landcover_path,
    tile_size=512,
    stride=256,
    buffer_radius=0,
)

out_folder = "landcover" tiles = geoai.export_geotiff_tiles( in_raster=train_raster_path, out_folder=out_folder, in_class_data=train_landcover_path, tile_size=512, stride=256, buffer_radius=0, )

Train semantic segmentation model¶

Now we'll train a semantic segmentation model using the new train_object_detection function. This function supports various architectures from segmentation-models-pytorch:

• Architectures: unet, deeplabv3, deeplabv3plus, fpn, pspnet, linknet, manet
• Encoders: resnet34, resnet50, efficientnet-b0, mobilenet_v2, etc.

In [ ]:

# Train U-Net model
geoai.train_segmentation_model(
    images_dir=f"{out_folder}/images",
    labels_dir=f"{out_folder}/labels",
    output_dir=f"{out_folder}/unet_models",
    architecture="unet",
    encoder_name="resnet34",
    encoder_weights="imagenet",
    num_channels=4,
    num_classes=13,
    batch_size=8,
    num_epochs=50,
    learning_rate=0.001,
    val_split=0.2,
    verbose=True,
    plot_curves=True,
)

# Train U-Net model geoai.train_segmentation_model( images_dir=f"{out_folder}/images", labels_dir=f"{out_folder}/labels", output_dir=f"{out_folder}/unet_models", architecture="unet", encoder_name="resnet34", encoder_weights="imagenet", num_channels=4, num_classes=13, batch_size=8, num_epochs=50, learning_rate=0.001, val_split=0.2, verbose=True, plot_curves=True, )

Run inference¶

Now we'll use the trained model to make predictions on the test image.

In [ ]:

# Define paths
masks_path = "naip_test_semantic_prediction.tif"
model_path = f"{out_folder}/unet_models/best_model.pth"

# Define paths masks_path = "naip_test_semantic_prediction.tif" model_path = f"{out_folder}/unet_models/best_model.pth"

In [ ]:

# Run semantic segmentation inference
geoai.semantic_segmentation(
    input_path=test_raster_path,
    output_path=masks_path,
    model_path=model_path,
    architecture="unet",
    encoder_name="resnet34",
    num_channels=4,
    num_classes=13,
    window_size=512,
    overlap=256,
    batch_size=4,
)

# Run semantic segmentation inference geoai.semantic_segmentation( input_path=test_raster_path, output_path=masks_path, model_path=model_path, architecture="unet", encoder_name="resnet34", num_channels=4, num_classes=13, window_size=512, overlap=256, batch_size=4, )

Visualize results¶

In [ ]:

geoai.write_colormap(masks_path, train_landcover_path, output=masks_path)

geoai.write_colormap(masks_path, train_landcover_path, output=masks_path)

In [ ]:

geoai.view_raster(masks_path, basemap=test_raster_url)

geoai.view_raster(masks_path, basemap=test_raster_url)