Utilities Module

This module provides utility functions for the semantic segmentation pipeline.

It includes functions for analyzing segmentation maps, generating statistics, handling file operations, and setting up logging. These utilities are used throughout the segmentation pipeline to support various processing tasks.

get_segmentation_data_batch

get_segmentation_data_batch(segmentation_data, start, end)

Get a batch of segmentation data from the HDF5 file.

PARAMETER	DESCRIPTION
segmentation_data	TYPE: Dataset
start	Start index of the batch. TYPE: int
end	End index of the batch. TYPE: int

RETURNS	DESCRIPTION
ndarray	np.ndarray: A batch of segmentation data.

Source code in cityseg/utils.py

def get_segmentation_data_batch(
    segmentation_data: h5py.Dataset, start: int, end: int
) -> np.ndarray:
    """
    Get a batch of segmentation data from the HDF5 file.

    Args:
        segmentation_data:
        start (int): Start index of the batch.
        end (int): End index of the batch.

    Returns:
        np.ndarray: A batch of segmentation data.
    """
    return segmentation_data[start:end]

setup_logging

setup_logging(log_level, verbose=False)

Set up logging configuration for the application.

This function configures console and file logging with appropriate log levels and formats.

PARAMETER	DESCRIPTION
log_level	The log level for file logging (e.g., "INFO", "DEBUG"). TYPE: str
verbose	If True, set console logging to DEBUG level. TYPE: bool DEFAULT: False

Source code in cityseg/utils.py

def setup_logging(log_level: str, verbose: bool = False) -> None:
    """
    Set up logging configuration for the application.

    This function configures console and file logging with appropriate
    log levels and formats.

    Args:
        log_level (str): The log level for file logging (e.g., "INFO", "DEBUG").
        verbose (bool): If True, set console logging to DEBUG level.
    """
    logger.remove()  # Remove default handler

    # Determine console log level
    console_level = "DEBUG" if verbose else log_level

    # Console logging
    console_format = "<green>{time:YYYY-MM-DD HH:mm:ss}</green> | <level>{level: <8}</level> | <cyan>{name}</cyan>:<cyan>{function}</cyan>:<cyan>{line}</cyan> - <level>{message}</level>"
    logger.add(
        sys.stderr,
        format=console_format,
        level=console_level,
        colorize=True,
    )

    # File logging (always at INFO level or lower, JSON format)
    file_level = min(log_level, "INFO")
    logger.add(
        "segmentation.log",
        format="{message}",
        level=file_level,
        rotation="100 MB",
        retention="1 week",
        serialize=True,
    )

    logger.info(
        f"Logging initialized. Console level: {console_level}, File level: {file_level}"
    )

tqdm_context

tqdm_context(*args, **kwargs)

A context manager for tqdm progress bars.

This context manager ensures that the tqdm progress bar is properly initialized and closed, even if an exception occurs.

PARAMETER	DESCRIPTION
*args	Positional arguments to pass to tqdm. TYPE: Any DEFAULT: ()
**kwargs	Keyword arguments to pass to tqdm. TYPE: Any DEFAULT: {}

YIELDS	DESCRIPTION
tqdm	The tqdm progress bar object. TYPE:: tqdm

Source code in cityseg/utils.py

@contextmanager
def tqdm_context(*args: Any, **kwargs: Any) -> Iterator[tqdm]:
    """
    A context manager for tqdm progress bars.

    This context manager ensures that the tqdm progress bar is properly
    initialized and closed, even if an exception occurs.

    Args:
        *args: Positional arguments to pass to tqdm.
        **kwargs: Keyword arguments to pass to tqdm.

    Yields:
        tqdm: The tqdm progress bar object.
    """
    try:
        progress_bar = tqdm(*args, **kwargs)
        yield progress_bar
    finally:
        progress_bar.close()