open-mmlab/mmdetection · warning

checkpoint is None, use COCO classes by default.

Error message

checkpoint is None, use COCO classes by default.

What it means

Warning from init_detector: checkpoint is None so no weights are loaded (random init) and COCO class names are used for dataset_meta.

Source

Thrown at mmdet/apis/inference.py:70

    if isinstance(config, (str, Path)):
        config = Config.fromfile(config)
    elif not isinstance(config, Config):
        raise TypeError('config must be a filename or Config object, '
                        f'but got {type(config)}')
    if cfg_options is not None:
        config.merge_from_dict(cfg_options)
    elif 'init_cfg' in config.model.backbone:
        config.model.backbone.init_cfg = None

    scope = config.get('default_scope', 'mmdet')
    if scope is not None:
        init_default_scope(config.get('default_scope', 'mmdet'))

    model = MODELS.build(config.model)
    model = revert_sync_batchnorm(model)
    if checkpoint is None:
        warnings.simplefilter('once')
        warnings.warn('checkpoint is None, use COCO classes by default.')
        model.dataset_meta = {'classes': get_classes('coco')}
    else:
        checkpoint = load_checkpoint(model, checkpoint, map_location='cpu')
        # Weights converted from elsewhere may not have meta fields.
        checkpoint_meta = checkpoint.get('meta', {})

        # save the dataset_meta in the model for convenience
        if 'dataset_meta' in checkpoint_meta:
            # mmdet 3.x, all keys should be lowercase
            model.dataset_meta = {
                k.lower(): v
                for k, v in checkpoint_meta['dataset_meta'].items()
            }
        elif 'CLASSES' in checkpoint_meta:
            # < mmdet 3.x
            classes = checkpoint_meta['CLASSES']
            model.dataset_meta = {'classes': classes}
        else:

View on GitHub (pinned to cfd5d3a985)

Solutions

1. Pass a checkpoint path or URL: init_detector(cfg, 'faster_rcnn.pth')
2. If intentional (structure testing), ignore — inference results will be garbage

Example fix

// before
model = init_detector(cfg, None)
// after
model = init_detector(cfg, 'checkpoints/faster_rcnn_r50.pth')

assert checkpoint is not None, 'init_detector without checkpoint yields random weights'

When it happens

Trigger: Calling mmdet.apis.init_detector(config, checkpoint=None) (or omitting checkpoint).

Common situations: Building a model for config debugging / latency benchmarking and forgetting weights; also passing an empty-string checkpoint path.

Related errors

• Checkpoint is not loaded, and the inference result is calcul
• dataset_meta or class names are not saved in the checkpoint'
• dataset_meta or class names are not saved in the checkpoint'
• palette does not exist, random is used by default. You can a
• weights is None, use COCO classes by default.

AI-assisted analysis of open-mmlab/mmdetection@cfd5d3a985 (2026-08-27). Data as JSON: /api/errors/77170a7210c7306e. Report an issue: GitHub.