PaddlePaddle/PaddleOCR · error · RuntimeError

Unknown augmenter arg: {}

Error message

Unknown augmenter arg: {}

What it means

IAA (imgaug) augmenter construction in PaddleOCR maps each entry of the 'augmenter_args' list to an imgaug class via getattr(A, type). When the augmenter type string is not a recognized imgaug/Resize augmenter name, construction falls through and raises RuntimeError('Unknown augmenter arg: ' + str(args)).

Source

Thrown at ppocr/data/imaug/iaa_augment.py:130

            # Process individual transformation specified as dictionary
            augmenter_type = args["type"]
            augmenter_args = args.get("args", {})
            augmenter_args_mapped = self.map_arguments(augmenter_type, augmenter_args)
            augmenter_type_mapped = self.imgaug_to_albu.get(
                augmenter_type, augmenter_type
            )
            if augmenter_type_mapped == "Resize":
                return ImgaugLikeResize(**augmenter_args_mapped)
            else:
                cls = getattr(A, augmenter_type_mapped)
                return cls(
                    **{
                        k: self.to_tuple_if_list(v)
                        for k, v in augmenter_args_mapped.items()
                    }
                )
        else:
            raise RuntimeError("Unknown augmenter arg: " + str(args))

    # Map arguments to expected format for each augmenter type
    def map_arguments(self, augmenter_type, augmenter_args):
        augmenter_args = augmenter_args.copy()  # Avoid modifying the original arguments
        if augmenter_type == "Resize":
            # Ensure size is a valid 2-element list or tuple
            size = augmenter_args.get("size")
            if size:
                if not isinstance(size, (list, tuple)) or len(size) != 2:
                    raise ValueError(
                        f"'size' must be a list or tuple of two numbers, but got {size}"
                    )
                min_scale, max_scale = size
                return {
                    "scale_range": (min_scale, max_scale),
                    "interpolation": 1,  # Linear interpolation
                    "p": 1.0,
                }

View on GitHub (pinned to 2661c7c0ef)

Solutions

1. Check the exact spelling of the 'type' field against imgaug's augmenters module (python -c 'import imgaug.augmenters as A; print([n for n in dir(A) if n[0].isupper()])')
2. Use the supported special-case name 'Resize' for resize operations
3. Pin a compatible imgaug version (imgaug 0.4.0 is the usual requirement for PaddleOCR)
4. Print the args dict in the error and compare it field-by-field with a known-good config such as configs/det/det_db_iaa.yml style pipelines

Example fix

# before (config)
augmenter_args:
- { type: Fliplr, args: { p: 0.5 } }
- { type: Resise, args: { size: [640, 640] } }
# after
augmenter_args:
- { type: Fliplr, args: { p: 0.5 } }
- { type: Resize, args: { size: [640, 640] } }

import imgaug.augmenters as A
valid = set(n for n in dir(A) if n[0].isupper()) | {'Resize'}
for entry in augmenter_args:
    if entry['type'] not in valid:
        raise SystemExit(f"Unknown augmenter type {entry['type']}; valid: {sorted(valid)[:20]}...")

try:
    aug = IaaAugment(augmenter_args=augmenter_args)
except (RuntimeError, AttributeError, ValueError) as e:
    raise ValueError(f'Bad iaa augmenter_args {augmenter_args}: {e}') from e

When it happens

Trigger: Using the iaa_augment.IaaAugment pipeline with an augmenter_args entry whose 'type' key is misspelled (e.g. 'Fliplr' vs 'Fliplr', 'Affine', 'Resize') or names a class that does not exist in the imgaug library version installed (A here is the imgaug augmenters module or its shim).

Common situations: Custom yml configs that add imgaug augmenters not covered by the mapping table; upgrading/downgrading imgaug so a class name disappears; a typo in the type field such as 'Resise' or 'MotionBlur' (imgaug has no such class).

Related errors

• 'size' must be a list or tuple of two numbers, but got {size
• Asset "${assetName}" references unknown model asset "${asset
• RecResizeImg.image_shape is required in rec inference.yml
• Unexpected recognition channels: ${String(channels)}
• OCR pipeline config text must decode to an object.

AI-assisted analysis of PaddlePaddle/PaddleOCR@2661c7c0ef (2026-08-14). Data as JSON: /api/errors/0b2c9f505915a568. Report an issue: GitHub.