How to use x2paddle - 10 common examples

if not os.path.isfile(self.caffe_proto):
                raise Exception(
                    "The .py file compiled by caffe.proto is not exist.")
            (filepath,
             tempfilename) = os.path.split(os.path.abspath(self.caffe_proto))
            (filename, extension) = os.path.splitext(tempfilename)
            sys.path.append(filepath)
            out = __import__(filename)
        return out

    def import_caffe(self):
        self.caffepb = self.import_caffepb()
        self.NetParameter = self.caffepb.NetParameter


class CaffeGraphNode(GraphNode):
    def __init__(self, layer, type_str, layer_name=None):
        if layer_name is None:
            super(CaffeGraphNode,
                  self).__init__(layer,
                                 layer.name.replace('/', '_').replace('-', '_'))
        else:
            super(CaffeGraphNode,
                  self).__init__(layer,
                                 layer_name.replace('/', '_').replace('-', '_'))
        self.layer_type = type_str
        self.fluid_code = FluidCode()
        self.data = None

    def set_params(self, params):
        self.data = params

def build(self):
        for layer in self.model.node:
            self.node_map[layer.name.replace('/', '_').replace(
                '-', '_')] = TFGraphNode(layer, data_format=self.tf_data_format)

        for layer_name, node in self.node_map.items():
            for in_node in node.layer.input:
                in_node = in_node.replace('/',
                                          '_').replace('-',
                                                       '_').replace('^', '')
                if in_node not in self.node_map:
                    if in_node.strip().split(':')[0] in self.node_map:
                        self.connect(in_node.strip().split(':')[0], layer_name)
                    else:
                        raise Exception(
                            'input[{}] of node[{}] does not exist in node_map'.
                            format(in_node, layer_name))
                else:
                    self.connect(in_node, layer_name)

def caffe2paddle(proto, weight, save_dir, caffe_proto, params_merge=False):
    from x2paddle.decoder.caffe_decoder import CaffeDecoder
    from x2paddle.op_mapper.caffe_op_mapper import CaffeOpMapper
    from x2paddle.optimizer.caffe_optimizer import CaffeOptimizer
    import google.protobuf as gpb
    ver_part = gpb.__version__.split('.')
    version_satisfy = False
    if (int(ver_part[0]) == 3 and int(ver_part[1]) >= 6) \
        or (int(ver_part[0]) > 3):
        version_satisfy = True
    assert version_satisfy, 'google.protobuf >= 3.6.0 is required'
    print("Now translating model from caffe to paddle.")
    model = CaffeDecoder(proto, weight, caffe_proto)
    mapper = CaffeOpMapper(model)
    optimizer = CaffeOptimizer(mapper)
    optimizer.merge_bn_scale()
    optimizer.merge_op_activation()
    mapper.save_inference_model(save_dir, params_merge)

def caffe2paddle(proto, weight, save_dir, caffe_proto, params_merge=False):
    from x2paddle.decoder.caffe_decoder import CaffeDecoder
    from x2paddle.op_mapper.caffe_op_mapper import CaffeOpMapper
    from x2paddle.optimizer.caffe_optimizer import CaffeOptimizer
    import google.protobuf as gpb
    ver_part = gpb.__version__.split('.')
    version_satisfy = False
    if (int(ver_part[0]) == 3 and int(ver_part[1]) >= 6) \
        or (int(ver_part[0]) > 3):
        version_satisfy = True
    assert version_satisfy, 'google.protobuf >= 3.6.0 is required'
    print("Now translating model from caffe to paddle.")
    model = CaffeDecoder(proto, weight, caffe_proto)
    mapper = CaffeOpMapper(model)
    optimizer = CaffeOptimizer(mapper)
    optimizer.merge_bn_scale()
    optimizer.merge_op_activation()
    mapper.save_inference_model(save_dir, params_merge)

def caffe2paddle(proto, weight, save_dir, caffe_proto, params_merge=False):
    from x2paddle.decoder.caffe_decoder import CaffeDecoder
    from x2paddle.op_mapper.caffe_op_mapper import CaffeOpMapper
    from x2paddle.optimizer.caffe_optimizer import CaffeOptimizer
    import google.protobuf as gpb
    ver_part = gpb.__version__.split('.')
    version_satisfy = False
    if (int(ver_part[0]) == 3 and int(ver_part[1]) >= 6) \
        or (int(ver_part[0]) > 3):
        version_satisfy = True
    assert version_satisfy, 'google.protobuf >= 3.6.0 is required'
    print("Now translating model from caffe to paddle.")
    model = CaffeDecoder(proto, weight, caffe_proto)
    mapper = CaffeOpMapper(model)
    optimizer = CaffeOptimizer(mapper)
    optimizer.merge_bn_scale()
    optimizer.merge_op_activation()
    mapper.save_inference_model(save_dir, params_merge)

val_var = self.graph.get_input_node(node, idx=4, copy=True)

        self.omit_nodes.append(val_scale.layer_name)
        self.omit_nodes.append(val_b.layer_name)
        self.omit_nodes.append(val_mean.layer_name)
        self.omit_nodes.append(val_var.layer_name)

        momentum = node.get_attr('momentum', .9)
        epsilon = node.get_attr('epsilon', 1e-5)

        # Attribute: spatial is used in BatchNormalization-1,6,7
        spatial = bool(node.get_attr('spatial'))
        attr = {
            "momentum": momentum,
            "epsilon": epsilon,
            "data_layout": string('NCHW'),
            "is_test": True,
            "param_attr": string(val_scale.layer_name),
            "bias_attr": string(val_b.layer_name),
            "moving_mean_name": string(val_mean.layer_name),
            "moving_variance_name": string(val_var.layer_name),
            "use_global_stats": spatial,
            "name": string(node.layer_name)
        }
        node.fluid_code.add_layer("batch_norm",
                                  inputs=val_x,
                                  output=node,
                                  param_attr=attr)

if auto_pad == "SAME_UPPER" or auto_pad == "SAME_LOWER":
            input_shape = val_x.out_shapes[0]
            pad_h = get_same_padding(input_shape[2], kernel_shape[0],
                                     strides[0])
            pad_w = get_same_padding(input_shape[3], kernel_shape[1],
                                     strides[1])
            attr = {"paddings": pad_h + pad_w, "pad_value": 0.0}

        attr = {
            "pool_size": kernel_shape,
            "pool_type": string("max"),
            "pool_stride": strides,
            "pool_padding": paddings,
            "ceil_mode": ceil_mode,
            "name": string(node.layer_name),
            "exclusive": False
        }
        node.fluid_code.add_layer(fluid_op,
                                  inputs=val_x,
                                  output=node,
                                  param_attr=attr)

def create_parameter(self, node, parameter=None):
        if parameter is not None:
            node = parameter
        dtype = node.dtype
        shape = node.out_shapes[0]

        self.weights[node.layer_name] = node.weight
        attr = {
            'dtype': string(dtype),
            'shape': shape,
            'name': string(node.layer_name),
            'attr': string(node.layer_name),
            'default_initializer': 'Constant(0.0)'
        }
        node.fluid_code.add_layer("create_parameter",
                                  inputs=None,
                                  output=node,
                                  param_attr=attr)

def place_holder(self, node):
        self.input_shapes.append(node.out_shapes[0])
        attr = {
            "dtype": string(node.dtype),
            "shape": node.out_shapes[0],
            "name": string(node.layer_name),
            "append_batch_size": 'False'
        }

        node.fluid_code.add_layer("data",
                                  inputs=None,
                                  output=node,
                                  param_attr=attr)

elif len(pads) == 8:
            paddings = np.array(pads).reshape(
                (-1, 4)).transpose().flatten().tolist()  # SSEE -> SESE
            if sum(paddings[:4]) == 0:
                fluid_op = 'pad2d'
                paddings = paddings[4:]
                attr['mode'] = string(mode)
        attr['paddings'] = paddings
        if op_independent:
            attr['name'] = string(node.layer_name)
            node.fluid_code.add_layer(fluid_op,
                                      inputs=val_x,
                                      output=node,
                                      param_attr=attr)
        else:
            attr['name'] = string(node.layer_name + '_paded')
            node.fluid_code.add_layer(fluid_op,
                                      inputs=val_x,
                                      output=node.layer_name + '_paded',
                                      param_attr=attr)
            return node.layer_name + '_paded'