Whether the files in the tools/analysis_tools/get_flops.py have not been corrected in time in mmsegmentation 1.0.0+？

[LovingThresh]

the build function is:

model = build_segmentor(
        cfg.model,
        train_cfg=cfg.get('train_cfg'),
        test_cfg=cfg.get('test_cfg')).cuda()

where the train_cfg is 'dict(by_epoch=True, max_epochs=100, val_interval=1 )' and test_cfg is 'dict()'.

In the build_segmentation function:

def build_segmentor(cfg, train_cfg=None, test_cfg=None):
    """Build segmentor."""
    if train_cfg is not None or test_cfg is not None:
        warnings.warn(
            'train_cfg and test_cfg is deprecated, '
            'please specify them in model', UserWarning)
    assert cfg.get('train_cfg') is None or train_cfg is None, \
        'train_cfg specified in both outer field and model field '
    assert cfg.get('test_cfg') is None or test_cfg is None, \
        'test_cfg specified in both outer field and model field '
    return SEGMENTORS.build(
        cfg, default_args=dict(train_cfg=train_cfg, test_cfg=test_cfg))

will have a error:

AssertionError: train_cfg specified in both outer field and model field.

And this part is alse extra：

    if hasattr(model,  'forward_dummy'):
        model.forward = model.forward_dummy
    else:
        raise NotImplementedError(
            'FLOPs counter is currently not currently supported with {}'.format(model.__class__.__name__))

It is necessary to update.

[xiexinch]

#2706
It's in progress.

[xiexinch]

Hi @LovingThresh,
The pr #2706 has been merged.