Generate ONNX model via ONNX, input initialized

onnx_gen_model.py

# By eopXD (eopxd.com)
# Since it took me some time to figure it out, I guess this file can help out.
import onnx
from onnx import helper
from onnx import AttributeProto, TensorProto, GraphProto
from onnx import numpy_helper

xNP = np.array([[[[0., 1., 2.],  # (1, 1, 3, 3)
                [3., 4., 5.],
                [6., 7., 8.]]]]).astype(np.float32)
wNP = np.array([[[[1., 1., 1.],  # (1, 2, 3, 3)
                [1., 1., 1.],
                [1., 1., 1.]],
               [[1., 1., 1.],
                [1., 1., 1.],
                [1., 1., 1.]]]]).astype(np.float32)

bNP = np.array([0, 0])

yNP = np.array([[[[0., 1., 3., 3., 2.],  # (1, 2, 5, 5)
                [3., 8., 15., 12., 7.],
                [9., 21., 36., 27., 15.],
                [9., 20., 33., 24., 13.],
                [6., 13., 21., 15., 8.]],

               [[0., 1., 3., 3., 2.],
                [3., 8., 15., 12., 7.],
                [9., 21., 36., 27., 15.],
                [9., 20., 33., 24., 13.],
                [6., 13., 21., 15., 8.]]]]).astype(np.float32)

ConvTranspose = onnx.helper.make_node("ConvTranspose", ["X", "W", "B"], ["Y"])

xInit = numpy_helper.from_array(xNP, "X")
wInit = numpy_helper.from_array(wNP, "W")
bInit = numpy_helper.from_array(bNP, "B")

xTensor = helper.make_tensor_value_info('X', TensorProto.FLOAT, [1, 1, 3, 3])
wTensor = helper.make_tensor_value_info('W', TensorProto.FLOAT, [1, 2, 3, 3])
bTensor = helper.make_tensor_value_info('B', TensorProto.FLOAT, [2])
yTensor = helper.make_tensor_value_info('Y', TensorProto.FLOAT, [1, 2, 5, 5])

graph_def = helper.make_graph(
    [ConvTranspose],
    'test-model',
    [xTensor, wTensor, bTensor],
    [yTensor],
    [xInit, wInit, bInit]
)
model_def = helper.make_model(graph_def, producer_name='onnx-example')

print('The model is:\n{}'.format(model_def))
onnx.checker.check_model(model_def)
print('The model is checked!')
onnx.save(model_def, "SingleConvTranspose.onnx")