Python mxnet.ndarray.load() Examples

The following are 25 code examples of mxnet.ndarray.load(). You can vote up the ones you like or vote down the ones you don't like, and go to the original project or source file by following the links above each example. You may also want to check out all available functions/classes of the module mxnet.ndarray , or try the search function .
Example #1
Source File: utils.py    From dynamic-training-with-apache-mxnet-on-aws with Apache License 2.0 6 votes vote down vote up 
def load_params(dir_path="", epoch=None, name=""):
    prefix = os.path.join(dir_path, name)
    _, param_loading_path, _ = get_saving_path(prefix, epoch)
    while not os.path.isfile(param_loading_path):
        logging.info("in load_param, %s Not Found!" % param_loading_path)
        time.sleep(60)
    save_dict = nd.load(param_loading_path)
    arg_params = {}
    aux_params = {}
    for k, v in save_dict.items():
        tp, name = k.split(':', 1)
        if tp == 'arg':
            arg_params[name] = v
        if tp == 'aux':
            aux_params[name] = v
    return arg_params, aux_params, param_loading_path
Example #2
Source File: utils.py    From SNIPER-mxnet with Apache License 2.0 6 votes vote down vote up 
def load_params(dir_path="", epoch=None, name=""):
    prefix = os.path.join(dir_path, name)
    _, param_loading_path, _ = get_saving_path(prefix, epoch)
    while not os.path.isfile(param_loading_path):
        logging.info("in load_param, %s Not Found!" % param_loading_path)
        time.sleep(60)
    save_dict = nd.load(param_loading_path)
    arg_params = {}
    aux_params = {}
    for k, v in save_dict.items():
        tp, name = k.split(':', 1)
        if tp == 'arg':
            arg_params[name] = v
        if tp == 'aux':
            aux_params[name] = v
    return arg_params, aux_params, param_loading_path
Example #3
Source File: utils.py    From training_results_v0.6 with Apache License 2.0 6 votes vote down vote up 
def load_params(dir_path="", epoch=None, name=""):
    prefix = os.path.join(dir_path, name)
    _, param_loading_path, _ = get_saving_path(prefix, epoch)
    while not os.path.isfile(param_loading_path):
        logging.info("in load_param, %s Not Found!" % param_loading_path)
        time.sleep(60)
    save_dict = nd.load(param_loading_path)
    arg_params = {}
    aux_params = {}
    for k, v in save_dict.items():
        tp, name = k.split(':', 1)
        if tp == 'arg':
            arg_params[name] = v
        if tp == 'aux':
            aux_params[name] = v
    return arg_params, aux_params, param_loading_path
Example #4
Source File: utils.py    From dynamic-training-with-apache-mxnet-on-aws with Apache License 2.0 5 votes vote down vote up 
def load_misc(dir_path="", epoch=None, name=""):
    prefix = os.path.join(dir_path, name)
    _, _, misc_saving_path = get_saving_path(prefix, epoch)
    with open(misc_saving_path, 'r') as fp:
        misc = json.load(fp)
    return misc
Example #5
Source File: utils.py    From SNIPER-mxnet with Apache License 2.0 5 votes vote down vote up 
def load_npz(path):
    with numpy.load(path) as data:
        ret = {k: data[k] for k in data.keys()}
        return ret
Example #6
Source File: utils.py    From SNIPER-mxnet with Apache License 2.0 5 votes vote down vote up 
def load_misc(dir_path="", epoch=None, name=""):
    prefix = os.path.join(dir_path, name)
    _, _, misc_saving_path = get_saving_path(prefix, epoch)
    with open(misc_saving_path, 'r') as fp:
        misc = json.load(fp)
    return misc
Example #7
Source File: resnetv1b_pruned.py    From panoptic-fpn-gluon with Apache License 2.0 5 votes vote down vote up 
def resnet101_v1d_73(pretrained=False, root='~/.mxnet/models', ctx=cpu(0), **kwargs):
    """Constructs a ResNetV1d-101_2.2x model. Uses resnet101_v1d construction from resnetv1b.py

    Parameters
    ----------
    pretrained : bool or str
        Boolean value controls whether to load the default pretrained weights for model.
        String value represents the hashtag for a certain version of pretrained weights.
    root : str, default '~/.mxnet/models'
        Location for keeping the model parameters.
    ctx : Context, default CPU
        The context in which to load the pretrained weights.
    """
    model = ResNetV1b(BottleneckV1b, [3, 4, 23, 3], deep_stem=True, avg_down=True,
                      name_prefix='resnetv1d_', **kwargs)
    dirname = os.path.dirname(__file__)
    json_filename = os.path.join(dirname, 'resnet%d_v%dd_%.1fx' % (101, 1, 2.2) + ".json")
    with open(json_filename, "r") as jsonFile:
        params_shapes = json.load(jsonFile)
    if pretrained:
        from ..model_store import get_model_file
        params_file = get_model_file('resnet%d_v%dd_%.1fx' % (101, 1, 2.2), tag=pretrained,
                                     root=root)
        prune_gluon_block(model, model.name, params_shapes, params=ndarray.load(params_file),
                          pretrained=True, ctx=ctx)
    else:
        prune_gluon_block(model, model.name, params_shapes, params=None, pretrained=False, ctx=ctx)

    if pretrained:
        from ...data import ImageNet1kAttr
        attrib = ImageNet1kAttr()
        model.synset = attrib.synset
        model.classes = attrib.classes
        model.classes_long = attrib.classes_long
    return model
Example #8
Source File: resnetv1b_pruned.py    From panoptic-fpn-gluon with Apache License 2.0 5 votes vote down vote up 
def resnet50_v1d_11(pretrained=False, root='~/.mxnet/models', ctx=cpu(0), **kwargs):
    """Constructs a ResNetV1d-50_8.8x model. Uses resnet50_v1d construction from resnetv1b.py

    Parameters
    ----------
    pretrained : bool or str
        Boolean value controls whether to load the default pretrained weights for model.
        String value represents the hashtag for a certain version of pretrained weights.
    root : str, default '~/.mxnet/models'
        Location for keeping the model parameters.
    ctx : Context, default CPU
        The context in which to load the pretrained weights.
    """
    model = ResNetV1b(BottleneckV1b, [3, 4, 6, 3], deep_stem=True, avg_down=True,
                      name_prefix='resnetv1d_', **kwargs)
    dirname = os.path.dirname(__file__)
    json_filename = os.path.join(dirname, 'resnet%d_v%dd_%.1fx' % (50, 1, 8.8) + ".json")
    with open(json_filename, "r") as jsonFile:
        params_shapes = json.load(jsonFile)
    if pretrained:
        from ..model_store import get_model_file
        params_file = get_model_file('resnet%d_v%dd_%.1fx' % (50, 1, 8.8), tag=pretrained,
                                     root=root)
        prune_gluon_block(model, model.name, params_shapes, params=ndarray.load(params_file),
                          pretrained=True, ctx=ctx)
    else:
        prune_gluon_block(model, model.name, params_shapes, params=None, pretrained=False, ctx=ctx)

    if pretrained:
        from ...data import ImageNet1kAttr
        attrib = ImageNet1kAttr()
        model.synset = attrib.synset
        model.classes = attrib.classes
        model.classes_long = attrib.classes_long
    return model
Example #9
Source File: resnetv1b_pruned.py    From panoptic-fpn-gluon with Apache License 2.0 5 votes vote down vote up 
def resnet50_v1d_37(pretrained=False, root='~/.mxnet/models', ctx=cpu(0), **kwargs):
    """Constructs a ResNetV1d-50_5.9x model. Uses resnet50_v1d construction from resnetv1b.py

    Parameters
    ----------
    pretrained : bool or str
        Boolean value controls whether to load the default pretrained weights for model.
        String value represents the hashtag for a certain version of pretrained weights.
    root : str, default '~/.mxnet/models'
        Location for keeping the model parameters.
    ctx : Context, default CPU
        The context in which to load the pretrained weights.
    """
    model = ResNetV1b(BottleneckV1b, [3, 4, 6, 3], deep_stem=True, avg_down=True,
                      name_prefix='resnetv1d_', **kwargs)
    dirname = os.path.dirname(__file__)
    json_filename = os.path.join(dirname, 'resnet%d_v%dd_%.1fx' % (50, 1, 5.9) + ".json")
    with open(json_filename, "r") as jsonFile:
        params_shapes = json.load(jsonFile)
    if pretrained:
        from ..model_store import get_model_file
        params_file = get_model_file('resnet%d_v%dd_%.1fx' % (50, 1, 5.9), tag=pretrained,
                                     root=root)
        prune_gluon_block(model, model.name, params_shapes, params=ndarray.load(params_file),
                          pretrained=True, ctx=ctx)
    else:
        prune_gluon_block(model, model.name, params_shapes, params=None, pretrained=False, ctx=ctx)

    if pretrained:
        from ...data import ImageNet1kAttr
        attrib = ImageNet1kAttr()
        model.synset = attrib.synset
        model.classes = attrib.classes
        model.classes_long = attrib.classes_long
    return model
Example #10
Source File: resnetv1b_pruned.py    From panoptic-fpn-gluon with Apache License 2.0 5 votes vote down vote up 
def resnet50_v1d_48(pretrained=False, root='~/.mxnet/models', ctx=cpu(0), **kwargs):
    """Constructs a ResNetV1d-50_3.6x model. Uses resnet50_v1d construction from resnetv1b.py

    Parameters
    ----------
    pretrained : bool or str
        Boolean value controls whether to load the default pretrained weights for model.
        String value represents the hashtag for a certain version of pretrained weights.
    root : str, default '~/.mxnet/models'
        Location for keeping the model parameters.
    ctx : Context, default CPU
        The context in which to load the pretrained weights.
    """
    model = ResNetV1b(BottleneckV1b, [3, 4, 6, 3], deep_stem=True, avg_down=True,
                      name_prefix='resnetv1d_', **kwargs)
    dirname = os.path.dirname(__file__)
    json_filename = os.path.join(dirname, 'resnet%d_v%dd_%.1fx' % (50, 1, 3.6) + ".json")
    with open(json_filename, "r") as jsonFile:
        params_shapes = json.load(jsonFile)
    if pretrained:
        from ..model_store import get_model_file
        params_file = get_model_file('resnet%d_v%dd_%.1fx' % (50, 1, 3.6), tag=pretrained,
                                     root=root)
        prune_gluon_block(model, model.name, params_shapes, params=ndarray.load(params_file),
                          pretrained=True, ctx=ctx)
    else:
        prune_gluon_block(model, model.name, params_shapes, params=None, pretrained=False, ctx=ctx)

    if pretrained:
        from ...data import ImageNet1kAttr
        attrib = ImageNet1kAttr()
        model.synset = attrib.synset
        model.classes = attrib.classes
        model.classes_long = attrib.classes_long
    return model
Example #11
Source File: resnetv1b_pruned.py    From panoptic-fpn-gluon with Apache License 2.0 5 votes vote down vote up 
def resnet50_v1d_86(pretrained=False, root='~/.mxnet/models', ctx=cpu(0), **kwargs):
    """Constructs a ResNetV1d-50_1.8x model. Uses resnet50_v1d construction from resnetv1b.py

    Parameters
    ----------
    pretrained : bool or str
        Boolean value controls whether to load the default pretrained weights for model.
        String value represents the hashtag for a certain version of pretrained weights.
    root : str, default '~/.mxnet/models'
        Location for keeping the model parameters.
    ctx : Context, default CPU
        The context in which to load the pretrained weights.
    """
    model = ResNetV1b(BottleneckV1b, [3, 4, 6, 3], deep_stem=True, avg_down=True,
                      name_prefix='resnetv1d_', **kwargs)
    dirname = os.path.dirname(__file__)
    json_filename = os.path.join(dirname, 'resnet%d_v%dd_%.1fx' % (50, 1, 1.8) + ".json")
    with open(json_filename, "r") as jsonFile:
        params_shapes = json.load(jsonFile)
    if pretrained:
        from ..model_store import get_model_file
        params_file = get_model_file('resnet%d_v%dd_%.1fx' % (50, 1, 1.8), tag=pretrained,
                                     root=root)
        prune_gluon_block(model, model.name, params_shapes, params=ndarray.load(params_file),
                          pretrained=True, ctx=ctx)
    else:
        prune_gluon_block(model, model.name, params_shapes, params=None, pretrained=False, ctx=ctx)

    if pretrained:
        from ...data import ImageNet1kAttr
        attrib = ImageNet1kAttr()
        model.synset = attrib.synset
        model.classes = attrib.classes
        model.classes_long = attrib.classes_long
    return model
Example #12
Source File: resnetv1b_pruned.py    From panoptic-fpn-gluon with Apache License 2.0 5 votes vote down vote up 
def resnet18_v1b_89(pretrained=False, root='~/.mxnet/models', ctx=cpu(0), **kwargs):
    """Constructs a ResNetV1b-18_2.6x model. Uses resnet18_v1b construction from resnetv1b.py

    Parameters
    ----------
    pretrained : bool or str
        Boolean value controls whether to load the default pretrained weights for model.
        String value represents the hashtag for a certain version of pretrained weights.
    root : str, default '~/.mxnet/models'
        Location for keeping the model parameters.
    ctx : Context, default CPU
        The context in which to load the pretrained weights.
    """
    model = ResNetV1b(BasicBlockV1b, [2, 2, 2, 2], name_prefix='resnetv1b_', **kwargs)
    dirname = os.path.dirname(__file__)
    json_filename = os.path.join(dirname, 'resnet%d_v%db_%.1fx' % (18, 1, 2.6) + ".json")
    with open(json_filename, "r") as jsonFile:
        params_shapes = json.load(jsonFile)
    if pretrained:
        from ..model_store import get_model_file
        params_file = get_model_file('resnet%d_v%db_%.1fx' % (18, 1, 2.6), tag=pretrained,
                                     root=root)
        prune_gluon_block(model, model.name, params_shapes, params=ndarray.load(params_file),
                          pretrained=True, ctx=ctx)
    else:
        prune_gluon_block(model, model.name, params_shapes, params=None, pretrained=False, ctx=ctx)
    if pretrained:
        from ...data import ImageNet1kAttr
        attrib = ImageNet1kAttr()
        model.synset = attrib.synset
        model.classes = attrib.classes
        model.classes_long = attrib.classes_long
    return model
Example #13
Source File: utils.py    From training_results_v0.6 with Apache License 2.0 5 votes vote down vote up 
def load_npz(path):
    with numpy.load(path) as data:
        ret = {k: data[k] for k in data.keys()}
        return ret
Example #14
Source File: utils.py    From training_results_v0.6 with Apache License 2.0 5 votes vote down vote up 
def load_misc(dir_path="", epoch=None, name=""):
    prefix = os.path.join(dir_path, name)
    _, _, misc_saving_path = get_saving_path(prefix, epoch)
    with open(misc_saving_path, 'r') as fp:
        misc = json.load(fp)
    return misc
Example #15
Source File: oth_alpha_pose.py    From imgclsmob with MIT License 5 votes vote down vote up 
def _try_load_parameters(self, filename=None, model=None, ctx=None, allow_missing=False,
                         ignore_extra=False):
    def getblock(parent, name):
        if len(name) == 1:
            if name[0].isnumeric():
                return parent[int(name[0])]
            else:
                return getattr(parent, name[0])
        else:
            if name[0].isnumeric():
                return getblock(parent[int(name[0])], name[1:])
            else:
                return getblock(getattr(parent, name[0]), name[1:])
    if filename is not None:
        loaded = ndarray.load(filename)
    else:
        loaded = {k: v.data() for k, v in model._collect_params_with_prefix().items()}
    params = self._collect_params_with_prefix()
    if not loaded and not params:
        return

    if not any('.' in i for i in loaded.keys()):
        # legacy loading
        del loaded
        self.collect_params().load(
            filename, ctx, allow_missing, ignore_extra, self.prefix)
        return

    for name in loaded:
        if name in params:
            if params[name].shape != loaded[name].shape:
                continue
            params[name]._load_init(loaded[name], ctx)
Example #16
Source File: resnetv1b_pruned.py    From gluon-cv with Apache License 2.0 5 votes vote down vote up 
def resnet101_v1d_73(pretrained=False, root='~/.mxnet/models', ctx=cpu(0), **kwargs):
    """Constructs a ResNetV1d-101_2.2x model. Uses resnet101_v1d construction from resnetv1b.py

    Parameters
    ----------
    pretrained : bool or str
        Boolean value controls whether to load the default pretrained weights for model.
        String value represents the hashtag for a certain version of pretrained weights.
    root : str, default '~/.mxnet/models'
        Location for keeping the model parameters.
    ctx : Context, default CPU
        The context in which to load the pretrained weights.
    """
    model = ResNetV1b(BottleneckV1b, [3, 4, 23, 3], deep_stem=True, avg_down=True,
                      name_prefix='resnetv1d_', **kwargs)
    dirname = os.path.dirname(__file__)
    json_filename = os.path.join(dirname, 'resnet%d_v%dd_%.1fx' % (101, 1, 2.2) + ".json")
    with open(json_filename, "r") as jsonFile:
        params_shapes = json.load(jsonFile)
    if pretrained:
        from ..model_store import get_model_file
        params_file = get_model_file('resnet%d_v%dd_%.1fx' % (101, 1, 2.2), tag=pretrained,
                                     root=root)
        prune_gluon_block(model, model.name, params_shapes, params=ndarray.load(params_file),
                          pretrained=True, ctx=ctx)
    else:
        prune_gluon_block(model, model.name, params_shapes, params=None, pretrained=False, ctx=ctx)

    if pretrained:
        from ...data import ImageNet1kAttr
        attrib = ImageNet1kAttr()
        model.synset = attrib.synset
        model.classes = attrib.classes
        model.classes_long = attrib.classes_long
    return model
Example #17
Source File: resnetv1b_pruned.py    From gluon-cv with Apache License 2.0 5 votes vote down vote up 
def resnet101_v1d_76(pretrained=False, root='~/.mxnet/models', ctx=cpu(0), **kwargs):
    """Constructs a ResNetV1d-101_1.9x model. Uses resnet101_v1d construction from resnetv1b.py

    Parameters
    ----------
    pretrained : bool or str
        Boolean value controls whether to load the default pretrained weights for model.
        String value represents the hashtag for a certain version of pretrained weights.
    root : str, default '~/.mxnet/models'
        Location for keeping the model parameters.
    ctx : Context, default CPU
        The context in which to load the pretrained weights.
    """
    model = ResNetV1b(BottleneckV1b, [3, 4, 23, 3], deep_stem=True, avg_down=True,
                      name_prefix='resnetv1d_', **kwargs)
    dirname = os.path.dirname(__file__)
    json_filename = os.path.join(dirname, 'resnet%d_v%dd_%.1fx' % (101, 1, 1.9) + ".json")
    with open(json_filename, "r") as jsonFile:
        params_shapes = json.load(jsonFile)
    if pretrained:
        from ..model_store import get_model_file
        params_file = get_model_file('resnet%d_v%dd_%.1fx' % (101, 1, 1.9), tag=pretrained,
                                     root=root)
        prune_gluon_block(model, model.name, params_shapes, params=ndarray.load(params_file),
                          pretrained=True, ctx=ctx)
    else:
        prune_gluon_block(model, model.name, params_shapes, params=None, pretrained=False, ctx=ctx)

    if pretrained:
        from ...data import ImageNet1kAttr
        attrib = ImageNet1kAttr()
        model.synset = attrib.synset
        model.classes = attrib.classes
        model.classes_long = attrib.classes_long
    return model
Example #18
Source File: resnetv1b_pruned.py    From gluon-cv with Apache License 2.0 5 votes vote down vote up 
def resnet50_v1d_11(pretrained=False, root='~/.mxnet/models', ctx=cpu(0), **kwargs):
    """Constructs a ResNetV1d-50_8.8x model. Uses resnet50_v1d construction from resnetv1b.py

    Parameters
    ----------
    pretrained : bool or str
        Boolean value controls whether to load the default pretrained weights for model.
        String value represents the hashtag for a certain version of pretrained weights.
    root : str, default '~/.mxnet/models'
        Location for keeping the model parameters.
    ctx : Context, default CPU
        The context in which to load the pretrained weights.
    """
    model = ResNetV1b(BottleneckV1b, [3, 4, 6, 3], deep_stem=True, avg_down=True,
                      name_prefix='resnetv1d_', **kwargs)
    dirname = os.path.dirname(__file__)
    json_filename = os.path.join(dirname, 'resnet%d_v%dd_%.1fx' % (50, 1, 8.8) + ".json")
    with open(json_filename, "r") as jsonFile:
        params_shapes = json.load(jsonFile)
    if pretrained:
        from ..model_store import get_model_file
        params_file = get_model_file('resnet%d_v%dd_%.1fx' % (50, 1, 8.8), tag=pretrained,
                                     root=root)
        prune_gluon_block(model, model.name, params_shapes, params=ndarray.load(params_file),
                          pretrained=True, ctx=ctx)
    else:
        prune_gluon_block(model, model.name, params_shapes, params=None, pretrained=False, ctx=ctx)

    if pretrained:
        from ...data import ImageNet1kAttr
        attrib = ImageNet1kAttr()
        model.synset = attrib.synset
        model.classes = attrib.classes
        model.classes_long = attrib.classes_long
    return model
Example #19
Source File: resnetv1b_pruned.py    From gluon-cv with Apache License 2.0 5 votes vote down vote up 
def resnet50_v1d_48(pretrained=False, root='~/.mxnet/models', ctx=cpu(0), **kwargs):
    """Constructs a ResNetV1d-50_3.6x model. Uses resnet50_v1d construction from resnetv1b.py

    Parameters
    ----------
    pretrained : bool or str
        Boolean value controls whether to load the default pretrained weights for model.
        String value represents the hashtag for a certain version of pretrained weights.
    root : str, default '~/.mxnet/models'
        Location for keeping the model parameters.
    ctx : Context, default CPU
        The context in which to load the pretrained weights.
    """
    model = ResNetV1b(BottleneckV1b, [3, 4, 6, 3], deep_stem=True, avg_down=True,
                      name_prefix='resnetv1d_', **kwargs)
    dirname = os.path.dirname(__file__)
    json_filename = os.path.join(dirname, 'resnet%d_v%dd_%.1fx' % (50, 1, 3.6) + ".json")
    with open(json_filename, "r") as jsonFile:
        params_shapes = json.load(jsonFile)
    if pretrained:
        from ..model_store import get_model_file
        params_file = get_model_file('resnet%d_v%dd_%.1fx' % (50, 1, 3.6), tag=pretrained,
                                     root=root)
        prune_gluon_block(model, model.name, params_shapes, params=ndarray.load(params_file),
                          pretrained=True, ctx=ctx)
    else:
        prune_gluon_block(model, model.name, params_shapes, params=None, pretrained=False, ctx=ctx)

    if pretrained:
        from ...data import ImageNet1kAttr
        attrib = ImageNet1kAttr()
        model.synset = attrib.synset
        model.classes = attrib.classes
        model.classes_long = attrib.classes_long
    return model
Example #20
Source File: resnetv1b_pruned.py    From gluon-cv with Apache License 2.0 5 votes vote down vote up 
def resnet50_v1d_86(pretrained=False, root='~/.mxnet/models', ctx=cpu(0), **kwargs):
    """Constructs a ResNetV1d-50_1.8x model. Uses resnet50_v1d construction from resnetv1b.py

    Parameters
    ----------
    pretrained : bool or str
        Boolean value controls whether to load the default pretrained weights for model.
        String value represents the hashtag for a certain version of pretrained weights.
    root : str, default '~/.mxnet/models'
        Location for keeping the model parameters.
    ctx : Context, default CPU
        The context in which to load the pretrained weights.
    """
    model = ResNetV1b(BottleneckV1b, [3, 4, 6, 3], deep_stem=True, avg_down=True,
                      name_prefix='resnetv1d_', **kwargs)
    dirname = os.path.dirname(__file__)
    json_filename = os.path.join(dirname, 'resnet%d_v%dd_%.1fx' % (50, 1, 1.8) + ".json")
    with open(json_filename, "r") as jsonFile:
        params_shapes = json.load(jsonFile)
    if pretrained:
        from ..model_store import get_model_file
        params_file = get_model_file('resnet%d_v%dd_%.1fx' % (50, 1, 1.8), tag=pretrained,
                                     root=root)
        prune_gluon_block(model, model.name, params_shapes, params=ndarray.load(params_file),
                          pretrained=True, ctx=ctx)
    else:
        prune_gluon_block(model, model.name, params_shapes, params=None, pretrained=False, ctx=ctx)

    if pretrained:
        from ...data import ImageNet1kAttr
        attrib = ImageNet1kAttr()
        model.synset = attrib.synset
        model.classes = attrib.classes
        model.classes_long = attrib.classes_long
    return model
Example #21
Source File: resnetv1b_pruned.py    From gluon-cv with Apache License 2.0 5 votes vote down vote up 
def resnet18_v1b_89(pretrained=False, root='~/.mxnet/models', ctx=cpu(0), **kwargs):
    """Constructs a ResNetV1b-18_2.6x model. Uses resnet18_v1b construction from resnetv1b.py

    Parameters
    ----------
    pretrained : bool or str
        Boolean value controls whether to load the default pretrained weights for model.
        String value represents the hashtag for a certain version of pretrained weights.
    root : str, default '~/.mxnet/models'
        Location for keeping the model parameters.
    ctx : Context, default CPU
        The context in which to load the pretrained weights.
    """
    model = ResNetV1b(BasicBlockV1b, [2, 2, 2, 2], name_prefix='resnetv1b_', **kwargs)
    dirname = os.path.dirname(__file__)
    json_filename = os.path.join(dirname, 'resnet%d_v%db_%.1fx' % (18, 1, 2.6) + ".json")
    with open(json_filename, "r") as jsonFile:
        params_shapes = json.load(jsonFile)
    if pretrained:
        from ..model_store import get_model_file
        params_file = get_model_file('resnet%d_v%db_%.1fx' % (18, 1, 2.6), tag=pretrained,
                                     root=root)
        prune_gluon_block(model, model.name, params_shapes, params=ndarray.load(params_file),
                          pretrained=True, ctx=ctx)
    else:
        prune_gluon_block(model, model.name, params_shapes, params=None, pretrained=False, ctx=ctx)
    if pretrained:
        from ...data import ImageNet1kAttr
        attrib = ImageNet1kAttr()
        model.synset = attrib.synset
        model.classes = attrib.classes
        model.classes_long = attrib.classes_long
    return model
Example #22
Source File: utils.py    From gluon-cv with Apache License 2.0 5 votes vote down vote up 
def _load_from_pytorch(self, filename, ctx=None):
    import torch
    from mxnet import nd
    loaded = torch.load(filename)
    params = self._collect_params_with_prefix()

    new_params = {}

    for name in loaded:
        if 'bn' in name or 'batchnorm' in name or '.downsample.1.' in name:
            if 'weight' in name:
                mxnet_name = name.replace('weight', 'gamma')
            elif 'bias' in name:
                mxnet_name = name.replace('bias', 'beta')
            else:
                mxnet_name = name
            new_params[mxnet_name] = nd.array(loaded[name].cpu().data.numpy())
        else:
            new_params[name] = nd.array(loaded[name].cpu().data.numpy())

    for name in new_params:
        if name not in params:
            print('==={}==='.format(name))
            raise Exception
        if name in params:
            params[name]._load_init(new_params[name], ctx=ctx)
Example #23
Source File: utils.py    From gluon-cv with Apache License 2.0 5 votes vote down vote up 
def _try_load_parameters(self, filename=None, model=None, ctx=None, allow_missing=False,
                         ignore_extra=False):
    def getblock(parent, name):
        if len(name) == 1:
            if name[0].isnumeric():
                return parent[int(name[0])]
            else:
                return getattr(parent, name[0])
        else:
            if name[0].isnumeric():
                return getblock(parent[int(name[0])], name[1:])
            else:
                return getblock(getattr(parent, name[0]), name[1:])
    if filename is not None:
        loaded = ndarray.load(filename)
    else:
        loaded = {k: v.data() for k, v in model._collect_params_with_prefix().items()}
    params = self._collect_params_with_prefix()
    if not loaded and not params:
        return

    if not any('.' in i for i in loaded.keys()):
        # legacy loading
        del loaded
        self.collect_params().load(
            filename, ctx, allow_missing, ignore_extra, self.prefix)
        return

    for name in loaded:
        if name in params:
            if params[name].shape != loaded[name].shape:
                continue
            params[name]._load_init(loaded[name], ctx)
Example #24
Source File: utils.py    From dynamic-training-with-apache-mxnet-on-aws with Apache License 2.0 5 votes vote down vote up 
def load_npz(path):
    with numpy.load(path) as data:
        ret = {k: data[k] for k in data.keys()}
        return ret
Example #25
Source File: oth_alpha_pose.py    From imgclsmob with MIT License 4 votes vote down vote up 
def get_alphapose(name, dataset, num_joints, pretrained=False,
                  pretrained_base=False, ctx=mx.cpu(),
                  norm_layer=nn.BatchNorm, norm_kwargs=None,
                  root=os.path.join('~', '.mxnet', 'models'), **kwargs):
    r"""Utility function to return AlphaPose networks.

    Parameters
    ----------
    name : str
        Model name.
    dataset : str
        The name of dataset.
    pretrained : bool or str
        Boolean value controls whether to load the default pretrained weights for model.
        String value represents the hashtag for a certain version of pretrained weights.
    ctx : mxnet.Context
        Context such as mx.cpu(), mx.gpu(0).
    root : str
        Model weights storing path.

    Returns
    -------
    mxnet.gluon.HybridBlock
        The AlphaPose network.

    """
    if norm_kwargs is None:
        norm_kwargs = {}
    preact = FastSEResNet(name, norm_layer=norm_layer, **norm_kwargs)
    if not pretrained and pretrained_base:
        from gluoncv.model_zoo import get_model
        base_network = get_model(name, pretrained=True, root=root)
        _try_load_parameters(self=base_network, model=base_network)
    net = AlphaPose(preact, num_joints, **kwargs)
    if pretrained:
        from gluoncv.model_zoo.model_store import get_model_file
        full_name = '_'.join(('alpha_pose', name, dataset))
        net.load_parameters(get_model_file(full_name, tag=pretrained, root=root))
    else:
        import warnings
        with warnings.catch_warnings(record=True):
            warnings.simplefilter("always")
            net.collect_params().initialize()
    net.collect_params().reset_ctx(ctx)
    return net