open-mmlab/mmdetection · error · ValueError

the item in out_indices must in range(0, {len(self.layer_set

Error message

the item in out_indices must in range(0, {len(self.layer_setting)}). But received {index}

What it means

EfficientNet validates each entry of out_indices against range(0, len(layer_setting)) for the chosen arch; an index outside the number of stages raises this ValueError. out_indices selects which stage feature maps the backbone returns.

Source

Thrown at mmdet/models/backbones/efficientnet.py:280

                 act_cfg=dict(type='Swish'),
                 norm_eval=False,
                 with_cp=False,
                 init_cfg=[
                     dict(type='Kaiming', layer='Conv2d'),
                     dict(
                         type='Constant',
                         layer=['_BatchNorm', 'GroupNorm'],
                         val=1)
                 ]):
        super(EfficientNet, self).__init__(init_cfg)
        assert arch in self.arch_settings, \
            f'"{arch}" is not one of the arch_settings ' \
            f'({", ".join(self.arch_settings.keys())})'
        self.arch_setting = self.arch_settings[arch]
        self.layer_setting = self.layer_settings[arch[:1]]
        for index in out_indices:
            if index not in range(0, len(self.layer_setting)):
                raise ValueError('the item in out_indices must in '
                                 f'range(0, {len(self.layer_setting)}). '
                                 f'But received {index}')

        if frozen_stages not in range(len(self.layer_setting) + 1):
            raise ValueError('frozen_stages must be in range(0, '
                             f'{len(self.layer_setting) + 1}). '
                             f'But received {frozen_stages}')
        self.drop_path_rate = drop_path_rate
        self.out_indices = out_indices
        self.frozen_stages = frozen_stages
        self.conv_cfg = conv_cfg
        self.norm_cfg = norm_cfg
        self.act_cfg = act_cfg
        self.norm_eval = norm_eval
        self.with_cp = with_cp

        self.layer_setting = model_scaling(self.layer_setting,
                                           self.arch_setting)

View on GitHub (pinned to cfd5d3a985)

Solutions

1. Set out_indices to valid stage indices 0..num_stages-1 for the chosen arch (check layer_settings[arch[:1]] length in efficientnet.py)
2. Match the number of out_indices to the neck's in_channels list
3. Prefer explicit small indices like (1,2,3,4) only after confirming the arch supports them

Example fix

# before
backbone=dict(type='EfficientNet', arch='b0', out_indices=(1,2,3,4,5,6,7))
# after
backbone=dict(type='EfficientNet', arch='b0', out_indices=(1,2,3,4))

n = len(backbone.layer_setting)
assert all(isinstance(i, int) and 0 <= i < n for i in out_indices), f'out_indices must be in [0, {n})'

def valid_out_indices(indices, n_stages: int) -> bool:
    return all(isinstance(i, int) and 0 <= i < n_stages for i in indices)

When it happens

Trigger: Setting out_indices=(1,2,3,4) on an EfficientNet arch whose stage list is shorter (e.g. b0 has stages 0..6 in mmdet but some archs fewer); using out_indices values copied from another backbone; negative indices (not allowed here, range starts at 0).

Common situations: Swapping backbone type in a neck config without adjusting out_indices; FPN configs assuming 5 outputs (0..4) on EfficientNet variants with different stage counts.

Related errors

• frozen_stages must be in range(0, {len(self.layer_setting) +
• frozen_stages must be in range(-1, len(arch_setting) + 1). B
• frozen_stages must be in range(-1, len(arch_setting) + 1). B
• invalid depth {depth} for darknet
• out_indices must be a subset of range(0, 8). But received {o

AI-assisted analysis of open-mmlab/mmdetection@cfd5d3a985 (2026-08-27). Data as JSON: /api/errors/08352ecca576e584. Report an issue: GitHub.