open-mmlab/mmdetection · error · ValueError

Invalid upsample method {}, accepted methods are "deconv", "

Error message

Invalid upsample method {}, accepted methods are "deconv", "nearest", "bilinear", "carafe"

What it means

FCNMaskHead validates upsample_cfg['type'] at construction: only None, 'deconv', 'nearest', 'bilinear', 'carafe' are accepted. Any other upsample type string raises ValueError immediately.

Source

Thrown at mmdet/models/roi_heads/mask_heads/fcn_mask_head.py:54

                 conv_out_channels: int = 256,
                 num_classes: int = 80,
                 class_agnostic: int = False,
                 upsample_cfg: ConfigType = dict(
                     type='deconv', scale_factor=2),
                 conv_cfg: OptConfigType = None,
                 norm_cfg: OptConfigType = None,
                 predictor_cfg: ConfigType = dict(type='Conv'),
                 loss_mask: ConfigType = dict(
                     type='CrossEntropyLoss', use_mask=True, loss_weight=1.0),
                 init_cfg: OptMultiConfig = None) -> None:
        assert init_cfg is None, 'To prevent abnormal initialization ' \
                                 'behavior, init_cfg is not allowed to be set'
        super().__init__(init_cfg=init_cfg)
        self.upsample_cfg = upsample_cfg.copy()
        if self.upsample_cfg['type'] not in [
                None, 'deconv', 'nearest', 'bilinear', 'carafe'
        ]:
            raise ValueError(
                f'Invalid upsample method {self.upsample_cfg["type"]}, '
                'accepted methods are "deconv", "nearest", "bilinear", '
                '"carafe"')
        self.num_convs = num_convs
        # WARN: roi_feat_size is reserved and not used
        self.roi_feat_size = _pair(roi_feat_size)
        self.in_channels = in_channels
        self.conv_kernel_size = conv_kernel_size
        self.conv_out_channels = conv_out_channels
        self.upsample_method = self.upsample_cfg.get('type')
        self.scale_factor = self.upsample_cfg.pop('scale_factor', None)
        self.num_classes = num_classes
        self.class_agnostic = class_agnostic
        self.conv_cfg = conv_cfg
        self.norm_cfg = norm_cfg
        self.predictor_cfg = predictor_cfg
        self.loss_mask = MODELS.build(loss_mask)

View on GitHub (pinned to cfd5d3a985)

Solutions

1. Fix the typo: use 'deconv', 'nearest', 'bilinear', or 'carafe'
2. For transposed convolution include kernel_size etc.: dict(type='deconv', kernel_size=4, stride=2)
3. Set upsample_cfg=None (or omit) if no upsampling is wanted

Example fix

# before
upsample_cfg=dict(type='bilinear2', scale_factor=2)
# after
upsample_cfg=dict(type='bilinear', scale_factor=2.0)

VALID_UP = {None, 'deconv', 'nearest', 'bilinear', 'carafe'}
assert upsample_cfg is None or upsample_cfg['type'] in VALID_UP

def valid_upsample_cfg(c): return c is None or c.get('type') in {None,'deconv','nearest','bilinear','carafe'}

When it happens

Trigger: mask_head=dict(type='FCNMaskHead', upsample_cfg=dict(type='bubic'/'deconvolution'/'pixelshuffle', ...)) — a typo'd or unsupported upsampling method name.

Common situations: Hand-writing mask head configs for Mask R-CNN; using Detectron2/mmseg-style upsample names; setting upsample_cfg type that mmdet never supported.

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

• LoadImageFromFile is not found in the test pipeline
• Visualization needs the "visualizer" termdefined in the conf
• config must be a filename or Config object, but got {type(co
• Unrecognized dataset: {dataset}
• The `file_client_args` is deprecated, please use `backend_ar

AI-assisted analysis of open-mmlab/mmdetection@cfd5d3a985 (2026-08-27). Data as JSON: /api/errors/6afd00c3ac24efbf. Report an issue: GitHub.