Standalone Usage

Every component can be run individually outside a pipeline via its run_standalone() classmethod. Each method has a component-specific signature with explicit parameters, so you get clear parameter names and IDE autocomplete.

For details on available config parameters (tile size, NMS thresholds, score weights, etc.), see Configuration.

Tilerizer

from canopyrs.engine.components.tilerizer import TilerizerComponent
from canopyrs.engine.config_parsers import TilerizerConfig

result = TilerizerComponent.run_standalone(
    config=TilerizerConfig(tile_type='tile', tile_size=512, ...),
    imagery_path='./raster.tif',
    output_path='./output',
)
print(result.tiles_path)

Detector

from canopyrs.engine.components.detector import DetectorComponent
from canopyrs.engine.config_parsers import DetectorConfig

result = DetectorComponent.run_standalone(
    config=DetectorConfig(model='dino_detrex', ...),
    tiles_path='./tiles',
    output_path='./output',
)
print(result.infer_gdf)

Using a config from a preset (see Model Zoo for available models):

config = DetectorConfig.from_yaml('canopyrs/config/detectors/dino_swinL_multi_NQOS.yaml')

result = DetectorComponent.run_standalone(
    config=config,
    tiles_path='./tiles',
    output_path='./output',
)

Segmenter

from canopyrs.engine.components.segmenter import SegmenterComponent
from canopyrs.engine.config_parsers import SegmenterConfig

result = SegmenterComponent.run_standalone(
    config=SegmenterConfig(model='sam3', ...),
    tiles_path='./tiles',
    output_path='./output',
    infer_coco_path='./coco.json',  # only if model requires box prompts
)
print(result.infer_gdf)

Using a config from a preset (see Model Zoo for available models):

config = SegmenterConfig.from_yaml('canopyrs/config/segmenters/sam3_multi_selvamask_FT.yaml')

result = SegmenterComponent.run_standalone(
    config=config,
    tiles_path='./tiles',
    output_path='./output',
    infer_coco_path='./coco.json',
)

Aggregator

from canopyrs.engine.components.aggregator import AggregatorComponent
from canopyrs.engine.config_parsers import AggregatorConfig

result = AggregatorComponent.run_standalone(
    config=AggregatorConfig(nms_threshold=0.5, ...),
    infer_gdf=my_detections_gdf,
    output_path='./output',
)
print(result.infer_gdf)

Classifier

from canopyrs.engine.components.classifier import ClassifierComponent
from canopyrs.engine.config_parsers import ClassifierConfig

result = ClassifierComponent.run_standalone(
    config=ClassifierConfig(model='resnet50', ...),
    tiles_path='./polygon_tiles',
    infer_coco_path='./coco.json',
    output_path='./output',
)
print(result.infer_gdf)

How it works

Under the hood, run_standalone() delegates to the generic run_component() helper, which wraps the component in a single-component pipeline. Inputs are validated before execution — if something is missing, you get a clear error message listing what's needed.

If you need more control, you can use run_component() directly:

from canopyrs.engine.pipeline import run_component
from canopyrs.engine.components.detector import DetectorComponent

result = run_component(
    component=DetectorComponent(config),
    output_path='./output',
    tiles_path='./tiles',
)