pyflink.multimodal.operators.image_remove_background#

image_remove_background(*columns, alpha_matting=False, alpha_matting_foreground_threshold=240, alpha_matting_background_threshold=10, alpha_matting_erode_size=10, bgcolor=None, model_sharing=None, concurrency=None, batch_size=None, num_gpus=None, gpu_type=None)[source]#

Remove the background from an image using rembg (U2-Net based).

Requires pip install rembg. This is a pandas batch UDF, but rembg inference is currently invoked per image inside each batch. num_gpus / gpu_type are forwarded as Flink resource hints; the rembg runtime decides whether the session actually uses GPU execution.

Parameters

*columns – Optional image columns. When provided, the UDF is applied directly instead of returning a factory.
alpha_matting – Enable alpha matting for smoother edge refinement. Default False.
alpha_matting_foreground_threshold – Foreground confidence threshold for alpha matting (0-255). Default 240.
alpha_matting_background_threshold – Background confidence threshold for alpha matting (0-255). Default 10.
alpha_matting_erode_size – Erosion kernel size for alpha matting refinement in pixels. Default 10.
bgcolor – Background fill color as (R, G, B, A) tuple. None (default) produces a transparent background.
model_sharing – Model sharing mode for the rembg session across parallel subtasks.
concurrency – UDF concurrency. None uses the framework default.
batch_size – Pandas batch size. None uses the framework default.
num_gpus – Fractional GPU count per subtask, e.g. 0.5. None runs on CPU.
gpu_type – Required GPU type, e.g. "A10". None accepts any available GPU.

Returns

A UDF producing decoded image values.

Example::

>>> # As a reusable variable
>>> remove_bg = image_remove_background(alpha_matting=True)
>>> df = df.with_columns("nobg", remove_bg(col("img")))
>>>
>>> # Inline
>>> df = df.with_columns("nobg", image_remove_background(col("img")))