def test_color_input(self, rank4_input_model, rank3_input_model):
mlmodel = ct.convert(
rank4_input_model,
inputs=[
ct.ImageType(shape=(1, 3, 10, 20),
color_layout=ct.colorlayout.RGB)
],
minimum_deployment_target=ct.target.macOS13,
)
assert_ops_in_mil_program(mlmodel,
expected_op_list=["cast", "add", "cast"])
assert_spec_input_image_type(
mlmodel._spec, expected_feature_type=ft.ImageFeatureType.RGB)
assert_prog_input_type(mlmodel._mil_program, expected_dtype_str="fp32")
assert_prog_output_type(mlmodel._mil_program,
expected_dtype_str="fp32")
verify_prediction(mlmodel)
with pytest.raises(ValueError, match="must have rank 4"):
mlmodel = ct.convert(
rank3_input_model,
inputs=[
ct.ImageType(shape=(1, 10, 20),
color_layout=ct.colorlayout.RGB)
],
minimum_deployment_target=ct.target.macOS12,
)
def process(self, cast_output_file=None):
try:
import coremltools
except:
LOG.logE(
"You need to install coremltools package if you want to convert PyTorch to CoreML model. E.g. pip install --upgrade coremltools"
)
return
output_coreml_file = self.config.coreml_model_dir
if cast_output_file:
output_coreml_file = '{}/coreml__{}.mlmodel'.format(
self.config.output_dir, cast_output_file)
self.config.coreml_model_dir = output_coreml_file
LOG.logI(
"config.coreml_model_dir found, save coreml model to {}...".format(
self.config.coreml_model_dir))
model = self.config.script_model_dir
if self.config.script_model_dir is None:
model = self.config.trace_model_dir
#input mode
if self.config.coreml_input_type == 'image':
input = coremltools.ImageType(
name="input",
shape=tuple(self.config.sample.shape),
scale=self.config.coreml_scale,
color_layout=self.config.coreml_color_layout,
bias=[
self.config.coreml_blue_bias,
self.config.coreml_green_bias, self.config.coreml_red_bias
])
else:
input = coremltools.TensorType(name='input',
shape=tuple(
self.config.sample.shape))
#convert
coreml_model = coremltools.convert(
model=model,
inputs=[input],
classfier_config=self.config.coreml_classfier_config,
minimum_deployment_target=self.config.
coreml_minimum_deployment_target)
# Set feature descriptions (these show up as comments in XCode)
coreml_model.input_description["input"] = "Deepvac Model Input"
coreml_model.output_description[
"classLabel"] = "Most likely image category"
# Set model author name
coreml_model.author = '"DeepVAC'
# Set the license of the model
coreml_model.license = "Deepvac Lincense"
coreml_model.short_description = "Powered by DeepVAC"
# Set a version for the model
coreml_model.version = self.config.coreml_version if self.config.coreml_version else "1.0"
# Save the CoreML model
coreml_model.save(output_coreml_file)
def test_image_input_enumerated(self, convert_to):
if convert_to == "mlprogram" and ct.utils._macos_version() < (12, 0):
return
example_input = torch.rand(1, 3, 50, 50) * 255
traced_model = torch.jit.trace(TestConvModule().eval(), example_input)
input_shape = ct.EnumeratedShapes(shapes=[[1, 3, 25,
25], [1, 3, 50, 50],
[1, 3, 67, 67]],
default=[1, 3, 67, 67])
model = ct.convert(traced_model,
inputs=[ct.ImageType(shape=input_shape)],
convert_to=convert_to)
spec = model.get_spec()
assert spec.description.input[0].type.imageType.width == 67
assert spec.description.input[0].type.imageType.height == 67
assert len(spec.description.input[0].type.imageType.enumeratedSizes.
sizes) == 3
assert spec.description.input[0].type.imageType.enumeratedSizes.sizes[
0].width == 25
assert spec.description.input[0].type.imageType.enumeratedSizes.sizes[
0].height == 25
_assert_torch_coreml_output_shapes(model,
spec,
traced_model,
example_input,
is_image_input=True)
def test_grayscale_fp16_output_image(self, rank4_grayscale_input_model):
mlmodel = ct.convert(
rank4_grayscale_input_model,
inputs=[ct.TensorType(name="input", shape=(1, 1, 10, 20))],
outputs=[
ct.ImageType(name="output_image",
color_layout=ct.colorlayout.GRAYSCALE_FLOAT16)
],
minimum_deployment_target=ct.target.macOS13,
compute_precision=ct.precision.FLOAT32,
)
sample_input = np.random.randint(low=0, high=200,
size=(1, 1, 10,
20)).astype(np.float32)
model_output_pil_image = mlmodel.predict({"input": sample_input
})['output_image']
assert isinstance(model_output_pil_image, Image.Image)
assert model_output_pil_image.mode == "F"
model_output_as_numpy = np.array(model_output_pil_image)
reference_output = rank4_grayscale_input_model(
torch.from_numpy(sample_input)).detach().numpy()
reference_output = np.squeeze(reference_output)
np.testing.assert_allclose(reference_output,
model_output_as_numpy,
rtol=1e-2,
atol=1e-2)
def test_program_bgr(self):
"""
Input graph:
main(x: ImageType(color_layout="BGR", channel_first=True)) {
y1 = relu(x)
y2 = relu(x)
output = add(y1, y2)
} [output]
Output graph:
main(x: ImageType(channel_first=True)) {
y1 = relu(x)
y2 = relu(x)
output = add(y1, y2)
} [output]
"""
@mb.program(input_specs=[mb.TensorSpec(shape=(1, 3, 20, 20))])
def prog(x):
y1 = mb.relu(x=x)
y2 = mb.relu(x=x)
z = mb.add(x=y1, y=y2)
return z
prog.main_input_types = (ct.ImageType(name='x',
shape=[1, 3, 20, 20],
color_layout="BGR",
channel_first=True), )
prev_prog, prev_block, block = apply_pass_and_basic_check(
prog, "mil_backend::insert_image_preprocessing_ops")
assert get_op_types_in_program(prev_prog) == ["relu", "relu", "add"]
assert get_op_types_in_program(prog) == ["relu", "relu", "add"]
def export_coreml(model, im, file, int8, half, prefix=colorstr('CoreML:')):
# YOLOv5 CoreML export
try:
check_requirements(('coremltools',))
import coremltools as ct
LOGGER.info(f'\n{prefix} starting export with coremltools {ct.__version__}...')
f = file.with_suffix('.mlmodel')
ts = torch.jit.trace(model, im, strict=False) # TorchScript model
ct_model = ct.convert(ts, inputs=[ct.ImageType('image', shape=im.shape, scale=1 / 255, bias=[0, 0, 0])])
bits, mode = (8, 'kmeans_lut') if int8 else (16, 'linear') if half else (32, None)
if bits < 32:
if platform.system() == 'Darwin': # quantization only supported on macOS
with warnings.catch_warnings():
warnings.filterwarnings("ignore", category=DeprecationWarning) # suppress numpy==1.20 float warning
ct_model = ct.models.neural_network.quantization_utils.quantize_weights(ct_model, bits, mode)
else:
print(f'{prefix} quantization only supported on macOS, skipping...')
ct_model.save(f)
LOGGER.info(f'{prefix} export success, saved as {f} ({file_size(f):.1f} MB)')
return ct_model, f
except Exception as e:
LOGGER.info(f'\n{prefix} export failure: {e}')
return None, None
def export_coreml(model, im, file, prefix=colorstr('CoreML:')):
# YOLOv5 CoreML export
ct_model = None
try:
check_requirements(('coremltools', ))
import coremltools as ct
print(
f'\n{prefix} starting export with coremltools {ct.__version__}...')
f = file.with_suffix('.mlmodel')
model.train() # CoreML exports should be placed in model.train() mode
ts = torch.jit.trace(model, im, strict=False) # TorchScript model
ct_model = ct.convert(ts,
inputs=[
ct.ImageType('image',
shape=im.shape,
scale=1 / 255.0,
bias=[0, 0, 0])
])
ct_model.save(f)
print(f'{prefix} export success, saved as {f} ({file_size(f):.1f} MB)')
except Exception as e:
print(f'\n{prefix} export failure: {e}')
return ct_model
def test_mil_enumerated_image(self):
enumerated_shapes = tuple([(1, 3, 10, 10), (1, 3, 10, 20),
(1, 3, 10, 30)])
input_shape = [
ct.ImageType(name="x",
shape=ct.EnumeratedShapes(shapes=enumerated_shapes))
]
mlmodel = ct.convert(self.basic_network,
source="milinternal",
convert_to="mlprogram",
inputs=input_shape)
input_spec = mlmodel.get_spec().description.input
assert len(input_spec) == 1, "1 input expected, got {} instead".format(
len(input_spec))
assert input_spec[
0].name == "x", "input name in MLModel is {}, 'x' is expected".format(
input_spec[0].name)
assert input_spec[0].type.WhichOneof(
"Type") == "imageType", "Expected imageType, got {}".format(
input_spec[0].type.WhichOneof("Type"))
assert input_spec[0].type.imageType.WhichOneof(
"SizeFlexibility"
) == "enumeratedSizes", "Expected enumeratedShapes in ShapeFlexibility"
spec_H = input_spec[0].type.imageType.height
spec_W = input_spec[0].type.imageType.width
assert spec_H == 10 and spec_W == 10, "expected [H, W] == [10, 10], got [{}, {}] instead".format(
spec_H, spec_W)
spec_enumerated_shapes = set()
for enumerated in input_spec[0].type.imageType.enumeratedSizes.sizes:
spec_enumerated_shapes.add(
tuple([1, 3, enumerated.height, enumerated.width]))
assert spec_enumerated_shapes == set(
enumerated_shapes), "Enumerated shape mismatch"
def get_input(name: str, inp: dict):
shape = get_shape(inp)
if 'type' in inp and inp['type'].upper() == 'IMAGE':
return ct.ImageType(name=name,
shape=shape,
bias=inp.get('bias'),
scale=inp.get('scale'))
else:
return ct.TensorType(name=name, shape=shape)
def test_color_output(self, rank4_input_model, float32_input_model_add_op):
# check that an error is raised if the output shape is not of form (1, 3, H, W)
with pytest.raises(ValueError,
match="must have rank 4. Instead it has rank 2"):
ct.convert(float32_input_model_add_op,
inputs=[ct.TensorType(shape=(10, 20))],
outputs=[ct.ImageType(color_layout=ct.colorlayout.RGB)],
minimum_deployment_target=ct.target.macOS13)
mlmodel = ct.convert(
rank4_input_model,
inputs=[
ct.ImageType(shape=(1, 3, 10, 20),
color_layout=ct.colorlayout.BGR)
],
outputs=[ct.ImageType(color_layout=ct.colorlayout.RGB)],
minimum_deployment_target=ct.target.macOS13,
)
assert_ops_in_mil_program(mlmodel,
expected_op_list=["cast", "add", "cast"])
assert_spec_input_image_type(
mlmodel._spec, expected_feature_type=ft.ImageFeatureType.BGR)
assert_spec_output_image_type(
mlmodel._spec, expected_feature_type=ft.ImageFeatureType.RGB)
assert_prog_input_type(mlmodel._mil_program, expected_dtype_str="fp32")
assert_prog_output_type(mlmodel._mil_program,
expected_dtype_str="fp32")
verify_prediction(mlmodel)
# check neural network conversion
mlmodel = ct.convert(
rank4_input_model,
inputs=[
ct.ImageType(shape=(1, 3, 10, 20),
color_layout=ct.colorlayout.RGB)
],
outputs=[ct.ImageType(color_layout=ct.colorlayout.BGR)],
)
assert_ops_in_mil_program(mlmodel, expected_op_list=["add"])
assert_spec_input_image_type(
mlmodel._spec, expected_feature_type=ft.ImageFeatureType.RGB)
assert_spec_output_image_type(
mlmodel._spec, expected_feature_type=ft.ImageFeatureType.BGR)
verify_prediction(mlmodel)
def test_grayscale_output(self, rank4_grayscale_input_model, rank4_grayscale_input_model_with_channel_first_output):
# check that an error is raised if the output shape is not of form (1, 1, H, W)
with pytest.raises(ValueError, match="Shape of the Grayscale image output,"):
mlmodel = ct.convert(rank4_grayscale_input_model,
inputs=[ct.ImageType(color_layout=ct.colorlayout.GRAYSCALE)],
outputs=[ct.ImageType(color_layout=ct.colorlayout.GRAYSCALE)],
)
with pytest.raises(TypeError, match="float16 dtype for outputs is only supported for deployment target >= iOS16/macOS13"):
mlmodel = ct.convert(rank4_grayscale_input_model_with_channel_first_output,
outputs=[ct.ImageType(color_layout=ct.colorlayout.GRAYSCALE_FLOAT16)],
minimum_deployment_target=ct.target.macOS12,
)
mlmodel = ct.convert(rank4_grayscale_input_model_with_channel_first_output,
inputs=[ct.ImageType(color_layout=ct.colorlayout.GRAYSCALE)],
outputs=[ct.ImageType(color_layout=ct.colorlayout.GRAYSCALE)],
)
assert_ops_in_mil_program(mlmodel, expected_op_list=["add"])
assert_spec_input_image_type(mlmodel._spec, expected_feature_type=ft.ImageFeatureType.GRAYSCALE)
assert_spec_output_image_type(mlmodel._spec, expected_feature_type=ft.ImageFeatureType.GRAYSCALE)
verify_prediction(mlmodel)
mlmodel = ct.convert(rank4_grayscale_input_model_with_channel_first_output,
inputs=[ct.ImageType(color_layout=ct.colorlayout.GRAYSCALE_FLOAT16)],
outputs=[ct.ImageType(color_layout=ct.colorlayout.GRAYSCALE_FLOAT16)],
minimum_deployment_target=ct.target.macOS13,
)
assert_cast_ops_count(mlmodel, expected_count=0)
assert_spec_input_image_type(mlmodel._spec, expected_feature_type=ft.ImageFeatureType.GRAYSCALE_FLOAT16)
assert_spec_output_image_type(mlmodel._spec, expected_feature_type=ft.ImageFeatureType.GRAYSCALE_FLOAT16)
assert_prog_input_type(mlmodel._mil_program, expected_dtype_str="fp16")
assert_prog_output_type(mlmodel._mil_program, expected_dtype_str="fp16")
verify_prediction(mlmodel)
mlmodel = ct.convert(rank4_grayscale_input_model_with_channel_first_output,
inputs=[ct.ImageType(color_layout=ct.colorlayout.GRAYSCALE)],
outputs=[ct.ImageType(color_layout=ct.colorlayout.GRAYSCALE_FLOAT16)],
minimum_deployment_target=ct.target.macOS13,
)
assert_ops_in_mil_program(mlmodel, expected_op_list=["cast", "add"])
assert_spec_input_image_type(mlmodel._spec, expected_feature_type=ft.ImageFeatureType.GRAYSCALE)
assert_spec_output_image_type(mlmodel._spec, expected_feature_type=ft.ImageFeatureType.GRAYSCALE_FLOAT16)
assert_prog_input_type(mlmodel._mil_program, expected_dtype_str="fp32")
assert_prog_output_type(mlmodel._mil_program, expected_dtype_str="fp16")
verify_prediction(mlmodel)
def apply(model,
outfile,
*,
input_w=129,
input_h=97,
minimum_deployment_target='iOS14'):
assert coremltools is not None
image_size_warning(model.base_net.stride, input_w, input_h)
# configure: inplace-ops are not supported
openpifpaf.network.heads.CompositeField3.inplace_ops = False
openpifpaf.network.heads.CompositeField4.inplace_ops = False
dummy_input = torch.randn(1, 3, input_h, input_w)
with torch.no_grad():
traced_model = torch.jit.trace(model, dummy_input)
coreml_model = coremltools.convert(
traced_model,
inputs=[
coremltools.ImageType(name='image',
shape=dummy_input.shape,
bias=[-1.0, -1.0, -1.0],
scale=1.0 / 127.0)
],
# classifier_config = ct.ClassifierConfig(class_labels)
minimum_deployment_target=getattr(coremltools.target,
minimum_deployment_target),
)
# pylint: disable=protected-access
coremltools.models.utils.rename_feature(
coreml_model._spec, coreml_model._spec.description.output[0].name,
'cif_head')
coremltools.models.utils.rename_feature(
coreml_model._spec, coreml_model._spec.description.output[1].name,
'caf_head')
# Meta
coreml_model.input_description['image'] = 'Input image to be classified'
coreml_model.output_description['cif_head'] = 'Composite Intensity Field'
coreml_model.output_description['caf_head'] = 'Composite Association Field'
coreml_model.author = 'Kreiss, Bertoni, Alahi: Composite Fields for Human Pose Estimation'
coreml_model.license = 'Please see https://github.com/openpifpaf/openpifpaf'
coreml_model.short_description = 'Composite Fields for Human Pose Estimation'
coreml_model.version = openpifpaf.__version__
coreml_model.save(outfile)
# test predict
image_input = PIL.Image.new('RGB', (input_w, input_h))
test_predict = coreml_model.predict({'image': image_input})
print('!!!!!!!!', test_predict)
def test_color_output(self, rank4_input_model, rank4_input_model_with_channel_first_output):
# check that an error is raised if the output shape is not of form (1, 3, H, W)
with pytest.raises(ValueError, match="Shape of the RGB/BGR image output,"):
mlmodel = ct.convert(rank4_input_model,
inputs=[ct.ImageType(color_layout=ct.colorlayout.RGB)],
outputs=[ct.ImageType(color_layout=ct.colorlayout.RGB)],
minimum_deployment_target=ct.target.macOS13,
)
mlmodel = ct.convert(rank4_input_model_with_channel_first_output,
inputs=[ct.ImageType(color_layout=ct.colorlayout.BGR)],
outputs=[ct.ImageType(color_layout=ct.colorlayout.RGB)],
minimum_deployment_target=ct.target.macOS13,
)
assert_ops_in_mil_program(mlmodel, expected_op_list=["cast", "add", "cast"])
assert_spec_input_image_type(mlmodel._spec, expected_feature_type=ft.ImageFeatureType.BGR)
assert_spec_output_image_type(mlmodel._spec, expected_feature_type=ft.ImageFeatureType.RGB)
assert_prog_input_type(mlmodel._mil_program, expected_dtype_str="fp32")
assert_prog_output_type(mlmodel._mil_program, expected_dtype_str="fp32")
verify_prediction(mlmodel)
# check neural network conversion
mlmodel = ct.convert(rank4_input_model_with_channel_first_output,
inputs=[ct.ImageType(color_layout=ct.colorlayout.RGB)],
outputs=[ct.ImageType(color_layout=ct.colorlayout.BGR)],
)
assert_ops_in_mil_program(mlmodel, expected_op_list=["add"])
assert_spec_input_image_type(mlmodel._spec, expected_feature_type=ft.ImageFeatureType.RGB)
assert_spec_output_image_type(mlmodel._spec, expected_feature_type=ft.ImageFeatureType.BGR)
verify_prediction(mlmodel)
def test_torch_image_enumerated_shapes():
import torch
import torchvision
torch_model = torchvision.models.mobilenet_v2().features
torch_model.eval()
example_input = torch.rand(1, 3, 256, 256)
traced_model = torch.jit.trace(torch_model, example_input)
input_shapes = ct.EnumeratedShapes(shapes=[(1, 3, 256, 256), (1, 3, 224, 224)])
image_input = ct.ImageType(shape=input_shapes,
bias=[-1, -1, -1], scale=1 / 127)
model = ct.convert(traced_model, inputs=[image_input])
assert model is not None
spec = model.get_spec()
assert len(spec.description.input[0].type.imageType.enumeratedSizes.sizes) == 2
def test_tf2_image_enumerated_shapes():
import tensorflow as tf
keras_model = tf.keras.applications.MobileNetV2(
input_shape=(None, None, 3,),
classes=1000,
include_top=False,
)
input_shapes = ct.EnumeratedShapes(shapes=[(1, 192, 192, 3), (1, 224, 224, 3)])
image_input = ct.ImageType(shape=input_shapes,
bias=[-1,-1,-1], scale=1/127)
model = ct.convert(keras_model, inputs=[image_input])
assert model is not None
spec = model.get_spec()
assert len(spec.description.input[0].type.imageType.enumeratedSizes.sizes) == 2
def test_scale_bias_types(self, scale_type, bias_type):
"""
Input graph:
main(x: ImageType(color_layout="RGB", scale=2.0, bias=[1.0, 2.0, 3.0], channel_first=True)) {
y1 = relu(x)
y2 = relu(x)
output = add(y1, y2)
} [output]
Output graph:
main(x: ImageType(channel_first=True)) {
y = mul(x, scale)
y_bias = add(y, bias)
y1 = relu(y_bias)
y2 = relu(y_bias)
output = add(y1, y2)
} [output]
"""
@mb.program(input_specs=[mb.TensorSpec(shape=(1, 3, 20, 20))])
def prog(x):
y1 = mb.relu(x=x)
y2 = mb.relu(x=x)
z = mb.add(x=y1, y=y2)
return z
prog.main_input_types = (ct.ImageType(name='x',
shape=[1, 3, 20, 20],
scale=scale_type(2.0),
bias=np.array(
[1, 2, 3]).astype(bias_type),
color_layout="RGB",
channel_first=True), )
prev_prog, prev_block, block = apply_pass_and_basic_check(
prog, "mil_backend::insert_image_preprocessing_ops")
assert get_op_types_in_program(prev_prog) == ["relu", "relu", "add"]
assert get_op_types_in_program(prog) == [
"mul", "add", "relu", "relu", "add"
]
scale_op = prog.find_ops(op_type="mul", exactly_one=True)[0]
assert scale_op.y.dtype() == prog.functions["main"].inputs["x"].dtype()
add_op = prog.find_ops(op_type="add", exactly_one=False)[0]
assert add_op.y.dtype() == prog.functions["main"].inputs["x"].dtype()
def convert_to_core_ml():
image_input = ct.ImageType(shape=(
1,
224,
224,
3,
),
bias=[-1, -1, -1],
scale=1 / 127)
classifier_config = ct.ClassifierConfig(saved_labels_path)
ml_model = ct.convert(
model,
inputs=[image_input],
classifier_config=classifier_config,
)
ml_model.save(saved_model_path + "/model.mlmodel")
def test_program_grayscale_with_scale_bias(self):
"""
Input graph:
main(x: ImageType(scale=2.0, bias=2.0, color_layout="G", channel_first=True)) {
y1 = relu(x)
y2 = relu(x)
output = add(y1, y2)
} [output]
Output graph:
main(x: ImageType(channel_first=True)) {
y_scaled = mul(x, 2)
y = add(y_scaled, 2)
y1 = relu(y)
y2 = relu(y)
output = add(y1, y2)
} [output]
"""
@mb.program(input_specs=[mb.TensorSpec(shape=(1, 1, 20, 20))])
def prog(x):
y1 = mb.relu(x=x)
y2 = mb.relu(x=x)
z = mb.add(x=y1, y=y2)
return z
prog.main_input_types = (ct.ImageType(name='x',
shape=[1, 1, 20, 20],
scale=2.0,
bias=2.0,
color_layout="G",
channel_first=True), )
prev_prog, prev_block, block = apply_pass_and_basic_check(
prog, "mil_backend::insert_image_preprocessing_ops")
assert get_op_types_in_program(prev_prog) == ["relu", "relu", "add"]
assert get_op_types_in_program(prog) == [
"mul", "add", "relu", "relu", "add"
]
scale_op = prog.find_ops(op_type="mul", exactly_one=True)[0]
assert scale_op.y.val == 2.0
add_op = prog.find_ops(op_type="add", exactly_one=False)[0]
assert add_op.y.val == 2.0
def create_core_ml_model_file(keras_model, filename):
if COREML_FILE_FORMAT not in filename:
filename += COREML_FILE_FORMAT
start = time.time()
image_input = ct.ImageType(shape=(
1,
224,
224,
3,
),
bias=[-91.4953, -103.8827, -131.0912],
color_layout="BGR")
coreml_model = ct.convert(keras_model, inputs=[image_input])
write_metadata(coreml_model)
coreml_model.save(filename)
end = time.time()
print(f"{filename} took {end - start} seconds to create.")
def test():
torch_model = torchvision.models.mobilenet_v2(pretrained=True)
# Set the model in evaluation mode
# torch_model.eval()
# example_input = torch.rand(1, 3, 224, 224) # after test, will get 'size mismatch' error message with size 256x256
# traced_model = torch.jit.trace(torch_model, example_input)
fastpunct = FastPunct()
traced_model = fastpunct.model
traced_model.save('aaa.h5')
# Convert to Core ML using the Unified Conversion API
model = ct.convert(
traced_model,
'pytorch',
inputs=[ct.ImageType(name="input_1", shape=example_input.shape)
] # provide only if step 2 was performed
)
# Save model
model.save("MobileNetV2.mlmodel")
def test_image_input_rangedim(self, convert_to):
example_input = torch.rand(1, 3, 50, 50) * 255
traced_model = torch.jit.trace(TestConvModule().eval(), example_input)
input_shape = ct.Shape(shape=(1, 3, ct.RangeDim(25, 100, default=45),
ct.RangeDim(25, 100, default=45)))
model = ct.convert(traced_model,
inputs=[ct.ImageType(shape=input_shape)],
convert_to=convert_to)
spec = model.get_spec()
assert spec.description.input[0].type.imageType.width == 45
assert spec.description.input[0].type.imageType.height == 45
assert spec.description.input[
0].type.imageType.imageSizeRange.widthRange.lowerBound == 25
assert spec.description.input[
0].type.imageType.imageSizeRange.widthRange.upperBound == 100
_assert_torch_coreml_output_shapes(model,
spec,
traced_model,
example_input,
is_image_input=True)
def test_grayscale_input_image(self, rank4_grayscale_input_model):
mlmodel = ct.convert(
rank4_grayscale_input_model,
inputs=[
ct.ImageType(name="input_image",
shape=(1, 1, 10, 20),
color_layout=ct.colorlayout.GRAYSCALE)
],
outputs=[ct.TensorType(name="output")],
minimum_deployment_target=ct.target.macOS13,
)
sample_input = np.random.randint(low=0, high=246, size=(1, 1, 10, 20))
img_input = Image.fromarray(sample_input[0, 0, :, :].astype(np.uint8),
'L')
model_output = mlmodel.predict({"input_image": img_input})['output']
reference_output = rank4_grayscale_input_model(
torch.from_numpy(sample_input.astype(
np.float32))).detach().numpy()
np.testing.assert_allclose(reference_output,
model_output,
rtol=1e-2,
atol=1e-2)
def test_grayscale_input(self, rank4_input_model, rank3_input_model, rank4_grayscale_input_model):
with pytest.raises(ValueError, match="must have rank 4"):
mlmodel = ct.convert(rank3_input_model,
inputs=[ct.ImageType(color_layout=ct.colorlayout.GRAYSCALE)],
minimum_deployment_target=ct.target.macOS13,
)
# invalid shape
with pytest.raises(ValueError):
mlmodel = ct.convert(rank4_input_model,
inputs=[ct.ImageType(color_layout=ct.colorlayout.GRAYSCALE)],
minimum_deployment_target=ct.target.macOS13,
)
mlmodel = ct.convert(rank4_grayscale_input_model,
inputs=[ct.ImageType(color_layout=ct.colorlayout.GRAYSCALE)],
minimum_deployment_target=ct.target.macOS13,
)
assert_ops_in_mil_program(mlmodel, expected_op_list=["cast", "transpose", "add", "cast"])
assert_spec_input_image_type(mlmodel._spec, expected_feature_type=ft.ImageFeatureType.GRAYSCALE)
assert_prog_input_type(mlmodel._mil_program, expected_dtype_str="fp32")
assert_prog_output_type(mlmodel._mil_program, expected_dtype_str="fp32")
verify_prediction(mlmodel)
with pytest.raises(TypeError, match="float16 dtype for inputs is only supported for deployment target >= iOS16/macOS13"):
mlmodel = ct.convert(rank4_grayscale_input_model,
inputs=[ct.ImageType(color_layout=ct.colorlayout.GRAYSCALE_FLOAT16)],
minimum_deployment_target=ct.target.macOS12,
)
# test that grayscale_16 raises error when used with neural network
with pytest.raises(TypeError, match="float16 dtype for inputs is only supported for deployment target >= iOS16/macOS13"):
mlmodel = ct.convert(rank4_grayscale_input_model,
inputs=[ct.ImageType(color_layout=ct.colorlayout.GRAYSCALE_FLOAT16)],
)
mlmodel = ct.convert(rank4_grayscale_input_model,
inputs=[ct.ImageType(color_layout=ct.colorlayout.GRAYSCALE_FLOAT16)],
outputs=[ct.TensorType(dtype=np.float16)],
minimum_deployment_target=ct.target.macOS13,
)
assert_ops_in_mil_program(mlmodel, expected_op_list=["transpose", "add"])
assert_spec_input_image_type(mlmodel._spec, expected_feature_type=ft.ImageFeatureType.GRAYSCALE_FLOAT16)
assert_prog_input_type(mlmodel._mil_program, expected_dtype_str="fp16")
assert_output_dtype(mlmodel, expected_type_str="fp16")
verify_prediction(mlmodel)
def convertToCoremlSpec(torchScript, sampleInput):
"""
Converts a torchscript to a coreml model
"""
try:
print(f"Starting CoreML conversion with coremltools {ct.__version__}")
nnSpec = ct.convert(
torchScript,
inputs=[
ct.ImageType(
name="image",
shape=sampleInput.shape,
scale=1 / 255.0,
bias=[0, 0, 0],
)
],
).get_spec()
print(f"CoreML conversion success")
except Exception as e:
print(f"CoreML conversion failure: {e}")
return
return nnSpec
def test_mil_ranged_image_with_default(self):
input_shape = [
ct.ImageType(name="x",
shape=(1, 3, 10, ct.RangeDim(10, 30, default=20)))
]
mlmodel = ct.convert(self.basic_network,
source="milinternal",
convert_to="mlprogram",
inputs=input_shape)
input_spec = mlmodel.get_spec().description.input
assert len(input_spec) == 1, "1 input expected, got {} instead".format(
len(input_spec))
assert input_spec[
0].name == "x", "input name in MLModel is {}, 'x' is expected".format(
input_spec[0].name)
assert input_spec[0].type.WhichOneof(
"Type") == "imageType", "Expected imageType, got {}".format(
input_spec[0].type.WhichOneof("Type"))
assert input_spec[0].type.imageType.WhichOneof(
"SizeFlexibility"
) == "imageSizeRange", "Expected imageSizeRange in ShapeFlexibility"
spec_H = input_spec[0].type.imageType.height
spec_W = input_spec[0].type.imageType.width
assert spec_H == 10 and spec_W == 20, "expected [H, W] == [10, 20], got [{}, {}] instead".format(
spec_H, spec_W)
spec_H_range = [
input_spec[0].type.imageType.imageSizeRange.heightRange.lowerBound,
input_spec[0].type.imageType.imageSizeRange.heightRange.upperBound
]
spec_W_range = [
input_spec[0].type.imageType.imageSizeRange.widthRange.lowerBound,
input_spec[0].type.imageType.imageSizeRange.widthRange.upperBound
]
assert spec_H_range == [10, 10], "Ranged height mismatch"
assert spec_W_range == [10, 30], "Ranged width mismatch"
def model_conversion():
print('Running model conversion')
opt, model = load_model_with_options()
if opt.eval:
model.eval()
# Read image and create input tensor
input_tensor = create_normalized_tensor(opt.input_img)
# Model conversion
model.netG.eval()
# # Step 1 - create a traced model
traced_model = torch.jit.trace(model.netG, input_tensor)
if opt.core_input == 'image':
ssmodel = ct.convert(traced_model,
inputs=[
ct.ImageType(name="input1",
shape=input_tensor.shape,
bias=[-1, -1, -1],
scale=1 / 127.0)
])
ssmodel.save(opt.model_path)
# Test model
input_img = Image.open(opt.input_img)
res = ssmodel.predict({"input1": input_img})
elif opt.core_input == 'tensor':
ssmodel = ct.convert(
traced_model,
inputs=[ct.TensorType(name="input1", shape=input_tensor.shape)])
ssmodel.save(opt.model_path)
# Test model
res = ssmodel.predict({"input1": input_tensor.numpy()})
if opt.res_img != '':
write_clmodel_res(opt.res_img, res['226'])
def test_grayscale_fp16_input_image(self, rank4_grayscale_input_model):
mlmodel = ct.convert(
rank4_grayscale_input_model,
inputs=[
ct.ImageType(name="input_image",
shape=(1, 1, 10, 20),
color_layout=ct.colorlayout.GRAYSCALE_FLOAT16)
],
outputs=[ct.TensorType(name="output")],
minimum_deployment_target=ct.target.macOS13,
)
# incorrect way to do prediction
with pytest.raises(
TypeError,
match="must be of type PIL.Image.Image with mode=='F'",
):
sample_input = np.random.randint(low=0,
high=246,
size=(1, 1, 10, 20))
img_input = Image.fromarray(
sample_input[0, 0, :, :].astype(np.uint8), 'L')
mlmodel.predict({"input_image": img_input})
# correct way to do prediction
sample_input = np.random.rand(1, 1, 10, 20) # in between [0, 1]
img_input = Image.fromarray(
sample_input[0, 0, :, :].astype(np.float32), 'F')
model_output = mlmodel.predict({"input_image": img_input})['output']
reference_output = rank4_grayscale_input_model(
torch.from_numpy(sample_input.astype(
np.float32))).detach().numpy()
np.testing.assert_allclose(reference_output,
model_output,
rtol=1e-2,
atol=1e-2)
print(onnx.helper.printable_graph(
onnx_model.graph)) # print a human readable model
print('ONNX export success, saved as %s' % f)
except Exception as e:
print('ONNX export failure: %s' % e)
# CoreML export
try:
import coremltools as ct
print('\nStarting CoreML export with coremltools %s...' %
ct.__version__)
# convert model from torchscript and apply pixel scaling as per detect.py
model = ct.convert(ts,
inputs=[
ct.ImageType(name='image',
shape=img.shape,
scale=1 / 255.0,
bias=[0, 0, 0])
])
f = opt.weights.replace('.pt', '.mlmodel') # filename
model.save(f)
print('CoreML export success, saved as %s' % f)
except Exception as e:
print('CoreML export failure: %s' % e)
# Finish
print(
'\nExport complete (%.2fs). Visualize with https://github.com/lutzroeder/netron.'
% (time.time() - t))
print('\nStarting ONNX export with onnx %s...' % onnx.__version__)
f = opt.weights.replace('.pt', '.onnx') # filename
model.fuse() # only for ONNX
torch.onnx.export(model, img, f, verbose=False, opset_version=12, input_names=['images'],
output_names=['classes', 'boxes'] if y is None else ['output'])
# Checks
onnx_model = onnx.load(f) # load onnx model
onnx.checker.check_model(onnx_model) # check onnx model
print(onnx.helper.printable_graph(onnx_model.graph)) # print a human readable model
print('ONNX export success, saved as %s' % f)
except Exception as e:
print('ONNX export failure: %s' % e)
# CoreML export
try:
import coremltools as ct
print('\nStarting CoreML export with coremltools %s...' % ct.__version__)
# convert model from torchscript and apply pixel scaling as per detect.py
model = ct.convert(ts, inputs=[ct.ImageType(name='images', shape=img.shape, scale=1 / 255.0, bias=[0, 0, 0])])
f = opt.weights.replace('.pt', '.mlmodel') # filename
model.save(f)
print('CoreML export success, saved as %s' % f)
except Exception as e:
print('CoreML export failure: %s' % e)
# Finish
print('\nExport complete. Visualize with https://github.com/lutzroeder/netron.')
def run(
weights='./yolov5s.pt', # weights path
img_size=(640, 640), # image (height, width)
batch_size=1, # batch size
device='cpu', # cuda device, i.e. 0 or 0,1,2,3 or cpu
include=('torchscript', 'onnx', 'coreml'), # include formats
half=False, # FP16 half-precision export
inplace=False, # set YOLOv5 Detect() inplace=True
train=False, # model.train() mode
optimize=False, # TorchScript: optimize for mobile
dynamic=False, # ONNX: dynamic axes
simplify=False, # ONNX: simplify model
opset_version=12, # ONNX: opset version
):
t = time.time()
include = [x.lower() for x in include]
img_size *= 2 if len(img_size) == 1 else 1 # expand
# Load PyTorch model
device = select_device(device)
assert not (
device.type == 'cpu' and opt.half
), '--half only compatible with GPU export, i.e. use --device 0'
model = attempt_load(weights, map_location=device) # load FP32 model
labels = model.names
# Input
gs = int(max(model.stride)) # grid size (max stride)
img_size = [check_img_size(x, gs)
for x in img_size] # verify img_size are gs-multiples
img = torch.zeros(batch_size, 3, *img_size).to(
device) # image size(1,3,320,192) iDetection
# Update model
if half:
img, model = img.half(), model.half() # to FP16
model.train() if train else model.eval(
) # training mode = no Detect() layer grid construction
for k, m in model.named_modules():
m._non_persistent_buffers_set = set() # pytorch 1.6.0 compatibility
if isinstance(m, Conv): # assign export-friendly activations
if isinstance(m.act, nn.Hardswish):
m.act = Hardswish()
elif isinstance(m.act, nn.SiLU):
m.act = SiLU()
elif isinstance(m, Detect):
m.inplace = inplace
m.onnx_dynamic = dynamic
# m.forward = m.forward_export # assign forward (optional)
for _ in range(2):
y = model(img) # dry runs
print(
f"\n{colorstr('PyTorch:')} starting from {weights} ({file_size(weights):.1f} MB)"
)
# TorchScript export -----------------------------------------------------------------------------------------------
if 'torchscript' in include or 'coreml' in include:
prefix = colorstr('TorchScript:')
try:
print(
f'\n{prefix} starting export with torch {torch.__version__}...'
)
f = weights.replace('.pt', '.torchscript.pt') # filename
ts = torch.jit.trace(model, img, strict=False)
(optimize_for_mobile(ts) if optimize else ts).save(f)
print(
f'{prefix} export success, saved as {f} ({file_size(f):.1f} MB)'
)
except Exception as e:
print(f'{prefix} export failure: {e}')
# ONNX export ------------------------------------------------------------------------------------------------------
if 'onnx' in include:
prefix = colorstr('ONNX:')
try:
import onnx
print(f'{prefix} starting export with onnx {onnx.__version__}...')
f = weights.replace('.pt', '.onnx') # filename
torch.onnx.export(
model,
img,
f,
verbose=False,
opset_version=opset_version,
training=torch.onnx.TrainingMode.TRAINING
if train else torch.onnx.TrainingMode.EVAL,
do_constant_folding=not train,
input_names=['images'],
output_names=['output'],
dynamic_axes={
'images': {
0: 'batch',
2: 'height',
3: 'width'
}, # shape(1,3,640,640)
'output': {
0: 'batch',
1: 'anchors'
} # shape(1,25200,85)
} if dynamic else None)
# Checks
model_onnx = onnx.load(f) # load onnx model
onnx.checker.check_model(model_onnx) # check onnx model
# print(onnx.helper.printable_graph(model_onnx.graph)) # print
# Simplify
if simplify:
try:
check_requirements(['onnx-simplifier'])
import onnxsim
print(
f'{prefix} simplifying with onnx-simplifier {onnxsim.__version__}...'
)
model_onnx, check = onnxsim.simplify(
model_onnx,
dynamic_input_shape=dynamic,
input_shapes={'images': list(img.shape)}
if dynamic else None)
assert check, 'assert check failed'
onnx.save(model_onnx, f)
except Exception as e:
print(f'{prefix} simplifier failure: {e}')
print(
f'{prefix} export success, saved as {f} ({file_size(f):.1f} MB)'
)
except Exception as e:
print(f'{prefix} export failure: {e}')
# CoreML export ----------------------------------------------------------------------------------------------------
if 'coreml' in include:
prefix = colorstr('CoreML:')
try:
import coremltools as ct
print(
f'{prefix} starting export with coremltools {ct.__version__}...'
)
assert train, 'CoreML exports should be placed in model.train() mode with `python export.py --train`'
model = ct.convert(ts,
inputs=[
ct.ImageType('image',
shape=img.shape,
scale=1 / 255.0,
bias=[0, 0, 0])
])
f = weights.replace('.pt', '.mlmodel') # filename
model.save(f)
print(
f'{prefix} export success, saved as {f} ({file_size(f):.1f} MB)'
)
except Exception as e:
print(f'{prefix} export failure: {e}')
# Finish
print(
f'\nExport complete ({time.time() - t:.2f}s). Visualize with https://github.com/lutzroeder/netron.'
)