def test_Reshape(self):
class Link(chainer.Chain):
def __call__(self, x):
return F.reshape(x, (-1,))
caffe.export(Link(), [self.x], None, True, 'test')
def main():
parser = argparse.ArgumentParser(description='Export')
parser.add_argument('--arch',
'-a',
type=str,
required=True,
choices=archs.keys(),
help='Arch name. models: ' + ', '.join(archs.keys()) +
'.')
parser.add_argument('--out-dir',
'-o',
type=str,
required=True,
help='Output directory name. '
'chainer_model.prototxt, chainer_model.caffemodel'
' will be created in it')
args = parser.parse_args()
if not os.path.exists(args.out_dir):
print('Created output directory: ' + args.out_dir)
os.mkdir(args.out_dir)
else:
print('Overwriting the existing directory: ' + args.out_dir)
if not os.path.isdir(args.out_dir):
raise ValueError(args.out_dir + ' exists but not a directory!')
print("load model")
model, input = get_network_for_imagenet(args.arch)
print("convert to caffe model")
caffe.export(model, [input], args.out_dir, True)
def main():
parser = argparse.ArgumentParser(description='Export')
parser.add_argument(
'--arch', '-a', type=str, required=True,
choices=archs.keys(),
help='Arch name. models: ' + ', '.join(archs.keys()) + '.')
parser.add_argument(
'--out-dir', '-o', type=str, required=True,
help='Output directory name. '
'chainer_model.prototxt, chainer_model.caffemodel'
' will be created in it')
args = parser.parse_args()
if not os.path.exists(args.out_dir):
print('Created output directory: ' + args.out_dir)
os.mkdir(args.out_dir)
else:
print('Overwriting the existing directory: ' + args.out_dir)
if not os.path.isdir(args.out_dir):
raise ValueError(args.out_dir + ' exists but not a directory!')
print("load model")
model, input = get_network_for_imagenet(args.arch)
print("convert to caffe model")
caffe.export(model, [input], args.out_dir, True)
def test_caffe_export_no_save(self):
class Model(chainer.Chain):
def __init__(self):
super(Model, self).__init__()
with self.init_scope():
self.l1 = L.Convolution2D(None, 1, 1, 1, 0)
self.b2 = L.BatchNormalization(1)
self.l3 = L.Linear(None, 1)
def __call__(self, x):
h = F.relu(self.l1(x))
h = self.b2(h)
return self.l3(h)
caffe.export(Model(), [self.x], None, True, 'test')
def test_caffe_export_no_save(self):
class Model(chainer.Chain):
def __init__(self):
super(Model, self).__init__()
with self.init_scope():
self.l1 = L.Convolution2D(None, 1, 1, 1, 0)
self.b2 = L.BatchNormalization(1)
self.l3 = L.Linear(None, 1)
def __call__(self, x):
h = F.relu(self.l1(x))
h = self.b2(h)
return self.l3(h)
caffe.export(Model(), [self.x], None, True, 'test')
def export_and_import(l, args, export_params=True, graph_name='test'):
"""Exports the given Link as Caffe model and returns the re-imported Link.
"""
with chainer.utils.tempdir() as tempdir:
caffe.export(l, args, tempdir, export_params, graph_name)
prototxt = os.path.join(tempdir, 'chainer_model.prototxt')
caffemodel = os.path.join(tempdir, 'chainer_model.caffemodel')
assert os.path.exists(prototxt)
assert os.path.exists(caffemodel)
# with open(prototxt) as f: print(f.read())
return L.caffe.CaffeFunction(caffemodel)
def export_and_import(l, args, export_params=True, graph_name='test'):
"""Exports the given Link as Caffe model and returns the re-imported Link.
"""
with chainer.utils.tempdir() as tempdir:
caffe.export(l, args, tempdir, export_params, graph_name)
prototxt = os.path.join(tempdir, 'chainer_model.prototxt')
caffemodel = os.path.join(tempdir, 'chainer_model.caffemodel')
assert os.path.exists(prototxt)
assert os.path.exists(caffemodel)
# with open(prototxt) as f: print(f.read())
return L.caffe.CaffeFunction(caffemodel)
def main():
for key, value in srcnn.archs.items():
model_dir = '../models/{}'.format(key.lower())
for filename in os.listdir(model_dir):
basename, ext = os.path.splitext(filename)
onnx_path = os.path.join(model_dir, basename + '.onnx')
if ext == '.npz':
model_path = os.path.join(model_dir, filename)
print(model_path)
channels = 3 if 'rgb' in filename else 1
model = value(channels)
size = 64 + model.offset
data = np.zeros((1, channels, size, size), dtype=np.float32)
x = chainer.Variable(data)
try:
chainer.serializers.load_npz(model_path, model)
caffe.export(model, [x], model_dir, True, basename)
rename_caffe_model(model_dir, basename)
except Exception:
print('Skipped caffe model export')
onnx_chainer.export(model, x, filename=onnx_path)
def rename_caffe_model(dir, filename):
model_path = os.path.join(dir, 'chainer_model.caffemodel')
prototxt_path = os.path.join(dir, 'chainer_model.prototxt')
new_model_path = os.path.join(dir, filename + '.caffemodel')
new_prototxt_path = os.path.join(dir, filename + '.prototxt')
os.rename(model_path, new_model_path)
os.rename(prototxt_path, new_prototxt_path)
if __name__ == '__main__':
for key, value in srcnn.archs.items():
model_dir = '../models/{}'.format(key.lower())
for filename in os.listdir(model_dir):
basename, ext = os.path.splitext(filename)
onnx_path = os.path.join(model_dir, basename + '.onnx')
if ext == '.npz':
model_path = os.path.join(model_dir, filename)
print(model_path)
channels = 3 if 'rgb' in filename else 1
model = value(channels)
size = 64 + model.offset
data = np.zeros((1, channels, size, size), dtype=np.float32)
x = chainer.Variable(data)
try:
chainer.serializers.load_npz(model_path, model)
caffe.export(model, [x], model_dir, True, basename)
rename_caffe_model(model_dir, basename)
except Exception:
print('Skipped caffe model export')
onnx_chainer.export(model, x, filename=onnx_path)
def test_Add(self):
class Link(chainer.Chain):
def __call__(self, x):
return x + x
caffe.export(Link(), [self.x], None, True, 'test')
def test_Softmax(self):
class Link(chainer.Chain):
def __call__(self, x):
return F.softmax(x)
caffe.export(Link(), [self.x], None, True, 'test')
def test_MaxPooling2D(self):
class Link(chainer.Chain):
def __call__(self, x):
return F.max_pooling_2d(x, 1, 1, 0)
caffe.export(Link(), [self.x], None, True, 'test')
def test_Reshape(self):
class Link(chainer.Chain):
def __call__(self, x):
return F.reshape(x, (-1, ))
caffe.export(Link(), [self.x], None, True, 'test')
def test_Add(self):
class Link(chainer.Chain):
def __call__(self, x):
return x + x
caffe.export(Link(), [self.x], None, True, 'test')
def test_Softmax(self):
class Link(chainer.Chain):
def __call__(self, x):
return F.softmax(x)
caffe.export(Link(), [self.x], None, True, 'test')
def test_MaxPooling2D(self):
class Link(chainer.Chain):
def __call__(self, x):
return F.max_pooling_2d(x, 1, 1, 0)
caffe.export(Link(), [self.x], None, True, 'test')
