open-mmlab/mmdetection · error · NotImplementedError

RandomCenterCropPad only support two testing pad mode:logica

Error message

RandomCenterCropPad only support two testing pad mode:logical-or and size_divisor.

What it means

RandomCenterCropPad's test-time padding accepts only two modes: 'logical' (target = w | pad_value, e.g. 32/64/127) or 'size_divisor'. Any other first element of test_pad_mode raises NotImplementedError in _test_aug.

Source

Thrown at mmdet/datasets/transforms/transforms.py:2126

        Args:
            results (dict): Image infomations in the augment pipeline.

        Returns:
            results (dict): The updated dict.
        """
        img = results['img']
        h, w, c = img.shape
        if self.test_pad_mode[0] in ['logical_or']:
            # self.test_pad_add_pix is only used for centernet
            target_h = (h | self.test_pad_mode[1]) + self.test_pad_add_pix
            target_w = (w | self.test_pad_mode[1]) + self.test_pad_add_pix
        elif self.test_pad_mode[0] in ['size_divisor']:
            divisor = self.test_pad_mode[1]
            target_h = int(np.ceil(h / divisor)) * divisor
            target_w = int(np.ceil(w / divisor)) * divisor
        else:
            raise NotImplementedError(
                'RandomCenterCropPad only support two testing pad mode:'
                'logical-or and size_divisor.')

        cropped_img, border, _ = self._crop_image_and_paste(
            img, [h // 2, w // 2], [target_h, target_w])
        results['img'] = cropped_img
        results['img_shape'] = cropped_img.shape[:2]
        results['border'] = border
        return results

    @autocast_box_type()
    def transform(self, results: dict) -> dict:
        img = results['img']
        assert img.dtype == np.float32, (
            'RandomCenterCropPad needs the input image of dtype np.float32,'
            ' please set "to_float32=True" in "LoadImageFromFile" pipeline')
        h, w, c = img.shape
        assert c == len(self.mean)

View on GitHub (pinned to cfd5d3a985)

Solutions

1. Use test_pad_mode=['logical', 32] (value must be a bitmask like 32/64/127)
2. Or use test_pad_mode=['size_divisor', 32]
3. Check exact spelling from the official YOLOX config

Example fix

# before
test_pad_mode=dict(type='logical', size=32)  # wrong structure
# after
test_pad_mode=['size_divisor', 32]

assert self_cfg['test_pad_mode'][0] in ('logical', 'size_divisor'), 'unsupported pad mode'

def valid_pad_mode(m):
    return isinstance(m, (list, tuple)) and len(m) == 2 and m[0] in ('logical', 'size_divisor')

When it happens

Trigger: Configuring dict(type='RandomCenterCropPad', test_pad_mode=['square', 32]) or a typo like 'logical_or' instead of 'logical', or passing a single value instead of a [mode, value] list.

Common situations: Customizing YOLOX test pipelines with a wrong pad mode name or malformed test_pad_mode tuple.

Understand the failure class

Background: "Invalid value" and "allowed values are" config errors: what your library rejected and how to fix it — this error's family across 41 libraries.

Related errors

• type must be a str or valid type, but got {type(obj_type)}
• RandomCenterCropPad only supports bbox.
• Not supported YouTubeVIS datasetversion: {dataset_version}
• NUM_BRANCHES({num_branches}) != NUM_BLOCKS({len(num_blocks)}
• NUM_BRANCHES({num_branches}) != NUM_CHANNELS({len(num_channe

AI-assisted analysis of open-mmlab/mmdetection@cfd5d3a985 (2026-08-27). Data as JSON: /api/errors/f2843b185c0c78ca. Report an issue: GitHub.