open-mmlab/mmdetection · info

palette does not exist, random is used by default. You can a

Error message

palette does not exist, random is used by default. You can also set the palette to customize.

What it means

Warning from init_detector: after checking the palette argument, config dataset metainfo, and checkpoint meta, no palette exists, so 'random' is used for visualization colors.

Source

Thrown at mmdet/apis/inference.py:108

            warnings.warn(
                'dataset_meta or class names are not saved in the '
                'checkpoint\'s meta data, use COCO classes by default.')
            model.dataset_meta = {'classes': get_classes('coco')}

    # Priority:  args.palette -> config -> checkpoint
    if palette != 'none':
        model.dataset_meta['palette'] = palette
    else:
        test_dataset_cfg = copy.deepcopy(config.test_dataloader.dataset)
        # lazy init. We only need the metainfo.
        test_dataset_cfg['lazy_init'] = True
        metainfo = DATASETS.build(test_dataset_cfg).metainfo
        cfg_palette = metainfo.get('palette', None)
        if cfg_palette is not None:
            model.dataset_meta['palette'] = cfg_palette
        else:
            if 'palette' not in model.dataset_meta:
                warnings.warn(
                    'palette does not exist, random is used by default. '
                    'You can also set the palette to customize.')
                model.dataset_meta['palette'] = 'random'

    model.cfg = config  # save the config in the model for convenience
    model.to(device)
    model.eval()
    return model


ImagesType = Union[str, np.ndarray, Sequence[str], Sequence[np.ndarray]]


def inference_detector(
    model: nn.Module,
    imgs: ImagesType,
    test_pipeline: Optional[Compose] = None,
    text_prompt: Optional[str] = None,

View on GitHub (pinned to cfd5d3a985)

Solutions

1. Pass palette: init_detector(cfg, ckpt, palette='coco')
2. Add 'palette' to dataset METAINFO in the config
3. Ignore if color scheme doesn't matter

Example fix

// before
model = init_detector(cfg, ckpt)  # random palette
// after
model = init_detector(cfg, ckpt, palette='coco')

if palette == 'none' and 'palette' not in getattr(model, 'dataset_meta', {}):
    model.dataset_meta['palette'] = 'coco'  # deterministic choice

When it happens

Trigger: init_detector(cfg, ckpt, palette='none') with a dataset/config/checkpoint that all lack a 'palette' entry.

Common situations: Custom datasets without palette in METAINFO; video inference via inference_detector then visualizer draws with random per-class colors.

Related errors

• palette does not exist, random is used by default. You can a
• Invalid type for palette: {type(palette)}
• checkpoint is None, use COCO classes by default.
• dataset_meta or class names are not saved in the checkpoint'
• The show is True, it means that only the prediction results

AI-assisted analysis of open-mmlab/mmdetection@cfd5d3a985 (2026-08-27). Data as JSON: /api/errors/7b37667d453437db. Report an issue: GitHub.