nemo_curator.stages.interleaved.filter.qrcode_filter

Module Contents

Classes

Name	Description
`InterleavedQRCodeFilterStage`	Filter interleaved image rows by QR code area ratio; drop images with high QR coverage.

Functions

Name	Description
`_qr_code_ratio`	Return the ratio of image area covered by all detected QR code(s), in [0, 1].

Data

DEFAULT_QRCODE_SCORE_THRESHOLD

API

class nemo_curator.stages.interleaved.filter.qrcode_filter.InterleavedQRCodeFilterStage(
    name: str = 'interleaved_qrcode_filter',
    drop_invalid_rows: bool = True,
    score_threshold: float = DEFAULT_QRCODE_SCORE_THRESHOLD
)

Dataclass

Bases: BaseInterleavedFilterStage

Filter interleaved image rows by QR code area ratio; drop images with high QR coverage.

name

str = 'interleaved_qrcode_filter'

score_threshold

float = DEFAULT_QRCODE_SCORE_THRESHOLD

nemo_curator.stages.interleaved.filter.qrcode_filter.InterleavedQRCodeFilterStage.content_keep_mask(
    task: nemo_curator.tasks.InterleavedBatch,
    df: pandas.DataFrame
) -> pandas.Series

nemo_curator.stages.interleaved.filter.qrcode_filter._qr_code_ratio(
    image: numpy.ndarray,
    row_index: collections.abc.Hashable | None = None
) -> float

Return the ratio of image area covered by all detected QR code(s), in [0, 1].

nemo_curator.stages.interleaved.filter.qrcode_filter.DEFAULT_QRCODE_SCORE_THRESHOLD: float = 0.05