CLI

dacapo

Command-line interface for the DACAPO application.

Args:

log_level (str): The desired log level for the application.

Examples:

To train a model, run: ` dacapo train --run-name my_run `

To validate a model, run: ` dacapo validate --run-name my_run --iteration 100 `

To apply a model, run: ` dacapo apply --run-name my_run --input-container /path/to/input --input-dataset my_dataset --output-path /path/to/output `

To predict with a model, run: ` dacapo predict --run-name my_run --iteration 100 --input-container /path/to/input --input-dataset my_dataset --output-path /path/to/output `

To run a blockwise operation, run: ` dacapo run-blockwise --input-container /path/to/input --input-dataset my_dataset --output-container /path/to/output --output-dataset my_output --worker-file /path/to/worker.py --total-roi [0:100,0:100,0:100] --read-roi-size [10,10,10] --write-roi-size [10,10,10] --num-workers 16 `

To segment blockwise, run: ` dacapo segment-blockwise --input-container /path/to/input --input-dataset my_dataset --output-container /path/to/output --output-dataset my_output --segment-function-file /path/to/segment_function.py --total-roi [0:100,0:100,0:100] --read-roi-size [10,10,10] --write-roi-size [10,10,10] --num-workers 16 `

dacapo [OPTIONS] COMMAND [ARGS]...

Options

--log-level <log_level>

Options:

DEBUG | INFO | WARNING | ERROR | CRITICAL

apply

Apply a trained run to an input dataset.

Args:

run_name (str): The name of the run to apply. input_container (Path | str): The path to the input container. input_dataset (str): The name of the input dataset. output_path (Path | str): The path to the output directory. validation_dataset (Dataset | str, optional): The name of the validation dataset. Defaults to None. criterion (str, optional): The criterion to use for applying the run. Defaults to “voi”. iteration (int, optional): The iteration of the model to use for prediction. Defaults to None. parameters (PostProcessorParameters | str, optional): The parameters for the post-processor. Defaults to None. roi (Roi | str, optional): The roi to predict on. Passed in as [lower:upper, lower:upper, … ]. Defaults to None. num_workers (int, optional): The number of workers to use for prediction. Defaults to 30. output_dtype (np.dtype | str, optional): The output data type. Defaults to “uint8”. overwrite (bool, optional): Whether to overwrite existing output files. Defaults to True.

Raises:

ValueError: If the run_name is not valid.

Examples:

To apply a trained run to an input dataset, run: ` dacapo apply --run-name my_run --input-container /path/to/input --input-dataset my_dataset --output-path /path/to/output `

dacapo apply [OPTIONS]

Options

-r, --run-name <run_name>

Required The name of the run to apply.

-ic, --input_container <input_container>

Required The path to the input container.

-id, --input_dataset <input_dataset>

Required The name of the input dataset.

-op, --output_path <output_path>

Required The path to the output directory.

-vd, --validation_dataset <validation_dataset>

The name of the validation dataset.

-c, --criterion <criterion>

The criterion to use for applying the run.

-i, --iteration <iteration>

The iteration of the model to use for prediction.

-p, --parameters <parameters>

The parameters for the post-processor.

-roi, --roi <roi>

The roi to predict on. Passed in as [lower:upper, lower:upper, … ]

-w, --num_workers <num_workers>

The number of workers to use for prediction.

-dt, --output_dtype <output_dtype>

The output data type.

-ow, --overwrite

Whether to overwrite existing output files.

config

dacapo config [OPTIONS]

predict

Apply a trained model to predict on a dataset.

Args:

run_name (str): The name of the run to apply. iteration (int): The training iteration of the model to use for prediction. input_container (Path | str): The path to the input container. input_dataset (str): The name of the input dataset. output_path (Path | str): The path to the output directory. output_roi (Optional[str | Roi], optional): The roi to predict on. Passed in as [lower:upper, lower:upper, … ]. Defaults to None. num_workers (int, optional): The number of workers to use for prediction. Defaults to 30. output_dtype (np.dtype | str, optional): The output data type. Defaults to np.uint8. overwrite (bool, optional): Whether to overwrite existing output files. Defaults to True.

Raises:

ValueError: If the run_name is not valid.

Examples:

To predict with a model, run: ` dacapo predict --run-name my_run --iteration 100 --input-container /path/to/input --input-dataset my_dataset --output-path /path/to/output `

dacapo predict [OPTIONS]

Options

-r, --run-name <run_name>

Required The name of the run to apply.

-i, --iteration <iteration>

Required The training iteration of the model to use for prediction.

-ic, --input_container <input_container>

Required The path to the input container.

-id, --input_dataset <input_dataset>

Required The name of the input dataset.

-op, --output_path <output_path>

Required The path to the output directory.

-roi, --output_roi <output_roi>

The roi to predict on. Passed in as [lower:upper, lower:upper, … ]

-w, --num_workers <num_workers>

The number of workers to use for prediction.

-dt, --output_dtype <output_dtype>

The output data type.

-ow, --overwrite

Whether to overwrite existing output files.

run-blockwise

Run blockwise processing on a dataset.

Args:

input_container: The path to the input container. input_dataset: The name of the input dataset. output_container: The path to the output container. output_dataset: The name of the output dataset. worker_file: The path to the worker file. total_roi: The total roi to be processed. Format is [start:end, start:end, … ] in voxels. Defaults to the roi of the input dataset. Do not use spaces in CLI argument. read_roi_size: The size of the roi to be read for each block, in the format of [z,y,x] in voxels. write_roi_size: The size of the roi to be written for each block, in the format of [z,y,x] in voxels. num_workers: The number of workers to use. max_retries: The maximum number of retries. timeout: The timeout in seconds. overwrite: Whether to overwrite existing output files. channels_out: The number of output channels. output_dtype: The output data type.

Raises:

ValueError: If the run_name is not valid.

Examples:

To run a blockwise operation, run: ` dacapo run-blockwise --input-container /path/to/input --input-dataset my_dataset --output-container /path/to/output --output-dataset my_output --worker-file /path/to/worker.py --total-roi [0:100,0:100,0:100] --read-roi-size [10,10,10] --write-roi-size [10,10,10] --num-workers 16 `

dacapo run-blockwise [OPTIONS]

Options

-ic, --input_container <input_container>

Required The path to the input container.

-id, --input_dataset <input_dataset>

Required The name of the input dataset.

-oc, --output_container <output_container>

Required The path to the output container.

-od, --output_dataset <output_dataset>

Required The name of the output dataset.

-w, --worker_file <worker_file>

Required The path to the worker file.

-tr, --total_roi <total_roi>

Required The total roi to be processed. Format is [start:end, start:end, … ] in voxels. Defaults to the roi of the input dataset. Do not use spaces in CLI argument.

-rr, --read_roi_size <read_roi_size>

Required The size of the roi to be read for each block, in the format of [z,y,x] in voxels.

-wr, --write_roi_size <write_roi_size>

Required The size of the roi to be written for each block, in the format of [z,y,x] in voxels.

-nw, --num_workers <num_workers>

The number of workers to use.

-mr, --max_retries <max_retries>

The maximum number of retries.

-t, --timeout <timeout>

The timeout in seconds.

-ow, --overwrite

Whether to overwrite existing output files.

-co, -channels_out <co>

The number of output channels.

-dt, --output_dtype <output_dtype>

The output data type.

segment-blockwise

Segment the input dataset blockwise using a segment function file.

Args:

input_container (str): The path to the input container. input_dataset (str): The name of the input dataset. output_container (str): The path to the output container. output_dataset (str): The name of the output dataset. segment_function_file (str): The path to the segment function file. total_roi (str): The total roi to be processed. Format is [start:end,start:end,…] in voxels. Defaults to the roi of the input dataset. Do not use spaces in CLI argument. read_roi_size (str): The size of the roi to be read for each block, in the format of [z,y,x] in voxels. write_roi_size (str): The size of the roi to be written for each block, in the format of [z,y,x] in voxels. context (str, optional): The context to be used, in the format of [z,y,x] in voxels. Defaults to the difference between the read and write rois. num_workers (int, optional): The number of workers to use. Defaults to 16. max_retries (int, optional): The maximum number of retries. Defaults to 2. timeout (int, optional): The timeout in seconds. Defaults to None. overwrite (bool, optional): Whether to overwrite existing output files. Defaults to True. channels_out (int, optional): The number of output channels. Defaults to None.

Raises:

ValueError: If the run_name is not valid.

Examples:

To segment blockwise, run: ` dacapo segment-blockwise --input-container /path/to/input --input-dataset my_dataset --output-container /path/to/output --output-dataset my_output --segment-function-file /path/to/segment_function.py --total-roi [0:100,0:100,0:100] --read-roi-size [10,10,10] --write-roi-size [10,10,10] --num-workers 16 `

dacapo segment-blockwise [OPTIONS]

Options

-ic, --input_container <input_container>

Required The path to the input container.

-id, --input_dataset <input_dataset>

Required The name of the input dataset.

-oc, --output_container <output_container>

Required The path to the output container.

-od, --output_dataset <output_dataset>

Required The name of the output dataset.

-sf, --segment_function_file <segment_function_file>

Required The path to the segment function file.

-tr, --total_roi <total_roi>

The total roi to be processed. Format is [start:end,start:end,…] in voxels. Defaults to the roi of the input dataset. Do not use spaces in CLI argument.

-rr, --read_roi_size <read_roi_size>

Required The size of the roi to be read for each block, in the format of [z,y,x] in voxels.

-wr, --write_roi_size <write_roi_size>

Required The size of the roi to be written for each block, in the format of [z,y,x] in voxels.

-c, --context <context>

The context to be used, in the format of [z,y,x] in voxels. Defaults to the difference between the read and write rois.

-nw, --num_workers <num_workers>

The number of workers to use.

-mr, --max_retries <max_retries>

The maximum number of retries.

-t, --timeout <timeout>

The timeout in seconds.

-ow, --overwrite

Whether to overwrite existing output files.

-co, --channels_out <channels_out>

The number of output channels.

train

Train a model with the specified run name.

Args:

run_name (str): The name of the run to train. no_validation (bool): Flag to disable validation after training.

dacapo train [OPTIONS]

Options

-r, --run-name <run_name>

Required The NAME of the run to train.

--no-validation

Disable validation after training.

validate

dacapo validate [OPTIONS]

Options

-r, --run-name <run_name>

Required The name of the run to validate.

-i, --iteration <iteration>

Required The iteration at which to validate the run.

-w, --num_workers <num_workers>

-dt, --output_dtype <output_dtype>

-ow, --overwrite