Implementation:Datajuicer Data juicer ImageAspectRatioFilter

Knowledge Sources	Datajuicer_Data_juicer
Domains	Data_Quality, Filtering
Last Updated	2026-02-14 16:00 GMT

Overview

Concrete tool for filtering data samples based on image aspect ratio provided by Data-Juicer.

Description

ImageAspectRatioFilter is a filter operator that keeps samples with image aspect ratio within a specific range. The aspect ratio is computed as width divided by height (W/H). The computed aspect ratios are cached under the aspect_ratios stats key. The operator supports 'any' (keep if any image meets the criteria) and 'all' (keep only if all images meet the criteria) strategies. It extends the Filter base class and implements the two-phase compute_stats/process pattern.

Usage

Import this operator when you need to filter dataset samples based on the aspect ratio of images. Configure it in your Data-Juicer YAML config or instantiate directly.

Code Reference

Source Location

• Repository: Datajuicer_Data_juicer
• File: data_juicer/ops/filter/image_aspect_ratio_filter.py
• Lines: 1-87

Signature

@OPERATORS.register_module("image_aspect_ratio_filter")
@LOADED_IMAGES.register_module("image_aspect_ratio_filter")
class ImageAspectRatioFilter(Filter):
    def __init__(self, min_ratio: float = 0.333, max_ratio: float = 3.0, any_or_all: str = "any", *args, **kwargs):
        ...

Import

from data_juicer.ops.filter.image_aspect_ratio_filter import ImageAspectRatioFilter

I/O Contract

Inputs

Name	Type	Required	Description
min_ratio	float	No	The minimum aspect ratio (W/H) to keep samples. Default: 0.333
max_ratio	float	No	The maximum aspect ratio (W/H) to keep samples. Default: 3.0
any_or_all	str	No	Keep strategy: 'any' or 'all' across images. Default: "any"

Outputs

Name	Type	Description
samples	Dict	Filtered samples with stats field updated (aspect_ratios)

Usage Examples

YAML Configuration

process:
  - image_aspect_ratio_filter:
      min_ratio: 0.333
      max_ratio: 3.0
      any_or_all: "any"

Python API

from data_juicer.ops.filter.image_aspect_ratio_filter import ImageAspectRatioFilter

op = ImageAspectRatioFilter(min_ratio=0.333, max_ratio=3.0)
# Apply to dataset
result = dataset.process(op)

Related Pages

• Environment:Datajuicer_Data_juicer_Python_Runtime_Environment