def _main():
parser = argparse.ArgumentParser()
parser.add_argument('-n', '--network', type=_text_type, help='Model Type', required=True,
choices = _default_model_info.keys())
parser.add_argument('-i', '--image', default=None,
type=_text_type, help='Test Image Path')
parser.add_argument('-o', '--output_dir', default='./',
type=_text_type, help='Caffe Checkpoint file name')
args = parser.parse_args()
if not download_file(_default_model_info[args.network]['prototxt'], directory=args.output_dir):
return -1
if not download_file(_default_model_info[args.network]['caffemodel'], directory=args.output_dir):
return -1
print("Model {} saved.".format(args.network))
if args.image:
# Yuhao TODO: inference code
pass
return 0
def download(cls, architecture, path='./'):
if cls.sanity_check(architecture):
cfg_name = architecture + ".cfg"
architecture_file = download_file(
cls.architecture_map[architecture]['config'],
directory=path,
local_fname=cfg_name)
if not architecture_file:
return None
weight_name = architecture + ".weights"
weight_file = download_file(
cls.architecture_map[architecture]['weights'],
directory=path,
local_fname=weight_name)
if not weight_file:
return None
print("Darknet Model {} saved as [{}] and [{}].".format(
architecture, architecture_file, weight_file))
return (architecture_file, weight_file)
else:
return None
def _main():
parser = argparse.ArgumentParser()
parser.add_argument('-n',
'--network',
type=_text_type,
help='Model Type',
required=True,
choices=DEFAULT_MODEL_INFO.keys())
parser.add_argument('-i',
'--image',
default=None,
type=_text_type,
help='Test Image Path')
parser.add_argument('-o',
'--output_dir',
default='./',
type=_text_type,
help='Caffe Checkpoint file name')
args = parser.parse_args()
arch_fn = download_file(DEFAULT_MODEL_INFO[args.network]['prototxt'],
directory=args.output_dir)
if not arch_fn:
return -1
weight_fn = download_file(DEFAULT_MODEL_INFO[args.network]['caffemodel'],
directory=args.output_dir)
if not weight_fn:
return -1
print("Model {} saved.".format(args.network))
if args.image:
import caffe
import numpy as np
from mmdnn.conversion.examples.imagenet_test import TestKit
net = caffe.Net(arch_fn, weight_fn, caffe.TEST)
func = TestKit.preprocess_func['caffe'][args.network]
img = func(args.image)
img = np.transpose(img, (2, 0, 1))
img = np.expand_dims(img, 0)
net.blobs['data'].data[...] = img
predict = np.squeeze(net.forward()['prob'][0])
predict = np.squeeze(predict)
top_indices = predict.argsort()[-5:][::-1]
result = [(i, predict[i]) for i in top_indices]
print(result)
print(np.sum(result))
return 0
def _main():
parser = argparse.ArgumentParser()
parser.add_argument('-n', '--network', type=_text_type, help='Model Type', required=True,
choices=_default_model_info.keys())
parser.add_argument('-i', '--image', default=None,
type=_text_type, help='Test Image Path')
parser.add_argument('-o', '--output_dir', default='./',
type=_text_type, help='Tensorflow Checkpoint file name')
args = parser.parse_args()
if not download_file(_default_model_info[args.network]['symbol'], directory=args.output_dir):
return -1
if not download_file(_default_model_info[args.network]['params'], directory=args.output_dir):
return -1
print("Model {} saved.".format(args.network))
file_name = _default_model_info[args.network]['params'].split('/')[-1]
prefix, epoch_num = file_name[:-7].rsplit('-', 1)
sym, arg_params, aux_params = mx.model.load_checkpoint(args.output_dir + prefix, int(epoch_num))
model = mx.mod.Module(symbol=sym)
model.bind(for_training=False,
data_shapes=[('data', (1, 3, _default_model_info[args.network]['image_size'],
_default_model_info[args.network]['image_size']))])
model.set_params(arg_params, aux_params, allow_missing=True, allow_extra=True)
if args.image:
import numpy as np
# need to be updated
network = _search_preprocess_key(args.network)
func = TestKit.preprocess_func['mxnet'][network]
img = func(args.image)
img = np.swapaxes(img, 0, 2)
img = np.swapaxes(img, 1, 2)
img = np.expand_dims(img, axis=0)
model.forward(Batch([mx.nd.array(img)]))
predict = model.get_outputs()[0].asnumpy()
predict = np.squeeze(predict)
top_indices = predict.argsort()[-5:][::-1]
result = [(i, predict[i]) for i in top_indices]
print(result)
return 0
def download(cls, architecture, path="./"):
if cls.sanity_check(architecture):
architecture_file = download_file(cls.architecture_map[architecture]['symbol'], directory=path)
if not architecture_file:
return None
weight_file = download_file(cls.architecture_map[architecture]['params'], directory=path)
if not weight_file:
return None
print("MXNet Model {} saved as [{}] and [{}].".format(architecture, architecture_file, weight_file))
return (architecture_file, weight_file)
else:
return None
def download(cls, architecture, path="./"):
if cls.sanity_check(architecture):
prototxt_name = architecture + "-deploy.prototxt"
architecture_file = download_file(cls.architecture_map[architecture]['prototxt'], directory=path ,local_fname=prototxt_name)
if not architecture_file:
return None
weight_file = download_file(cls.architecture_map[architecture]['caffemodel'], directory=path)
if not weight_file:
return None
print("Caffe Model {} saved as [{}] and [{}].".format(architecture, architecture_file, weight_file))
return (architecture_file, weight_file)
else:
return None
def download(cls, architecture, path="./"):
if cls.sanity_check(architecture):
architecture_file = download_file(
cls.architecture_map[architecture]['url'],
directory=path,
auto_unzip=True)
if not architecture_file:
return None
tf.reset_default_graph()
if 'ckpt' in cls.architecture_map[architecture]['filename']:
cls.handle_checkpoint(architecture, path)
elif cls.architecture_map[architecture]['filename'].endswith('pb'):
cls.handle_frozen_graph(architecture, path)
else:
raise ValueError("Unknown file name [{}].".format(
cls.architecture_map[architecture]['filename']))
return architecture_file
else:
return None
def download(cls, architecture, path="./"):
if cls.sanity_check(architecture):
prototxt_name = architecture + "-deploy.prototxt"
architecture_file = download_file(cls.architecture_map[architecture]['prototxt'], directory=path, local_fname=prototxt_name)
if not architecture_file:
return None
weight_name = architecture + ".caffemodel"
weight_file = download_file(cls.architecture_map[architecture]['caffemodel'], directory=path, local_fname=weight_name)
if not weight_file:
return None
print("Caffe Model {} saved as [{}] and [{}].".format(architecture, architecture_file, weight_file))
return (architecture_file, weight_file)
else:
return None
def inference(cls, architecture, files, model_path, image_path):
import numpy as np
if cls.sanity_check(architecture):
download_file(cls._base_model_url + "cfg/coco.data",
directory='./')
download_file(cls._base_model_url + "data/coco.names",
directory='./data/')
net = cdarknet.load_net(files[0], files[1], 0)
meta = cdarknet.load_meta("coco.data")
r = cdarknet.detect(net, meta, image_path)
# print(r)
return r
else:
return None
def _main():
parser = argparse.ArgumentParser()
parser.add_argument('-n', '--network', type=_text_type, help='Model Type', required=True,
choices=DEFAULT_MODEL_INFO.keys())
parser.add_argument('-i', '--image', default=None,
type=_text_type, help='Test Image Path')
parser.add_argument('-o', '--output_dir', default='./',
type=_text_type, help='Caffe Checkpoint file name')
args = parser.parse_args()
arch_fn = download_file(DEFAULT_MODEL_INFO[args.network]['prototxt'], directory=args.output_dir)
if not arch_fn:
return -1
weight_fn = download_file(DEFAULT_MODEL_INFO[args.network]['caffemodel'], directory=args.output_dir)
if not weight_fn:
return -1
print("Model {} saved.".format(args.network))
if args.image:
import caffe
import numpy as np
from mmdnn.conversion.examples.imagenet_test import TestKit
net = caffe.Net(arch_fn.encode("utf-8"), weight_fn.encode("utf-8"), caffe.TEST)
# net = caffe.Net(arch_fn, weight_fn, caffe.TEST)
func = TestKit.preprocess_func['caffe'][args.network]
img = func(args.image)
img = np.transpose(img, (2, 0, 1))
img = np.expand_dims(img, 0)
net.blobs['data'].data[...] = img
predict = np.squeeze(net.forward()['prob'][0])
predict = np.squeeze(predict)
top_indices = predict.argsort()[-5:][::-1]
result = [(i, predict[i]) for i in top_indices]
print(result)
print(np.sum(result))
return 0
def _main():
parser = argparse.ArgumentParser()
parser.add_argument('-n',
'--network',
type=_text_type,
help='Model Type',
required=True,
choices=MODEL_URL.keys())
parser.add_argument('-i',
'--image',
default=None,
type=_text_type,
help='Test Image Path')
parser.add_argument('-o',
'--output_dir',
default='./',
type=_text_type,
help='Caffe Checkpoint file name')
args = parser.parse_args()
fn = download_file(MODEL_URL[args.network], directory=args.output_dir)
if not fn:
return -1
model = C.Function.load(fn)
if len(model.outputs) > 1:
for idx, output in enumerate(model.outputs):
if len(output.shape) > 0:
eval_node = idx
break
model = C.as_composite(model[eval_node].owner)
model.save(fn)
print("Model {} is saved as {}.".format(args.network, fn))
if args.image:
import numpy as np
from mmdnn.conversion.examples.imagenet_test import TestKit
func = TestKit.preprocess_func['cntk'][args.network]
img = func(args.image)
img = np.transpose(img, (2, 0, 1))
predict = model.eval({model.arguments[0]: [img]})
predict = np.squeeze(predict)
top_indices = predict.argsort()[-5:][::-1]
result = [(i, predict[i]) for i in top_indices]
print(result)
print(np.sum(result))
return 0
def download(cls, architecture, path='./'):
if cls.sanity_check(architecture):
architecture_file = download_file(
cls.architecture_map[architecture], directory=path)
if not architecture_file:
return None
print('Coreml model {} is saved in [{}]'.format(
architecture, path))
return architecture_file
else:
return None
def download(cls, architecture, path="./"):
if cls.sanity_check(architecture):
architecture_file = download_file(cls.architecture_map[architecture], directory=path)
model = C.Function.load(architecture_file)
if len(model.outputs) > 1:
for idx, output in enumerate(model.outputs):
if len(output.shape) > 0:
eval_node = idx
break
model = C.as_composite(model[eval_node].owner)
model.save(architecture_file)
print("Cntk Model {} saved as [{}].".format(architecture, architecture_file))
return architecture_file
else:
return None
def _main():
parser = argparse.ArgumentParser()
parser.add_argument('-n', '--network', type=_text_type, help='Model Type', required=True,
choices=MODEL_URL.keys())
parser.add_argument('-i', '--image', default=None,
type=_text_type, help='Test Image Path')
parser.add_argument('-o', '--output_dir', default='./',
type=_text_type, help='Caffe Checkpoint file name')
args = parser.parse_args()
fn = download_file(MODEL_URL[args.network], directory=args.output_dir)
if not fn:
return -1
model = C.Function.load(fn)
if len(model.outputs) > 1:
for idx, output in enumerate(model.outputs):
if len(output.shape) > 0:
eval_node = idx
break
model = C.as_composite(model[eval_node].owner)
model.save(fn)
print("Model {} is saved as {}.".format(args.network, fn))
if args.image:
import numpy as np
from mmdnn.conversion.examples.imagenet_test import TestKit
func = TestKit.preprocess_func['cntk'][args.network]
img = func(args.image)
img = np.transpose(img, (2, 0, 1))
predict = model.eval({model.arguments[0]:[img]})
predict = np.squeeze(predict)
top_indices = predict.argsort()[-5:][::-1]
result = [(i, predict[i]) for i in top_indices]
print(result)
print(np.sum(result))
return 0
def download(cls, architecture, path="./"):
if cls.sanity_check(architecture):
architecture_file = download_file(cls.architecture_map[architecture]['url'], directory=path, auto_unzip=True)
if not architecture_file:
return None
if cls.architecture_map[architecture]['filename'].endswith('ckpt'):
cls.handle_checkpoint(architecture, path)
elif cls.architecture_map[architecture]['filename'].endswith('pb'):
cls.handle_frozen_graph(architecture, path)
else:
raise ValueError("Unknown file name [{}].".format(cls.architecture_map[architecture]['filename']))
return architecture_file
else:
return None
def download(cls, architecture, path="./"):
if cls.sanity_check(architecture):
architecture_file = download_file(
cls.architecture_map[architecture], directory=path)
model = C.Function.load(architecture_file)
if len(model.outputs) > 1:
for idx, output in enumerate(model.outputs):
if len(output.shape) > 0:
eval_node = idx
break
model = C.as_composite(model[eval_node].owner)
model.save(architecture_file)
print("Cntk Model {} saved as [{}].".format(
architecture, architecture_file))
return architecture_file
else:
return None
def download(cls, architecture, path="./"):
if cls.sanity_check(architecture):
reset_parser()
DATA_DIM = 3 * paddle_extractor._image_size * paddle_extractor._image_size # Use 3 * 331 * 331 or 3 * 299 * 299 for Inception-ResNet-v2.
CLASS_DIM = paddle_extractor.class_dim_map[architecture]
image = paddle.layer.data(
name="image", type=paddle.data_type.dense_vector(DATA_DIM))
if 'resnet' in architecture:
from mmdnn.conversion.examples.paddle.models import resnet
depth = int(architecture.strip('resnet'))
out = resnet.resnet_imagenet(image,
class_dim=CLASS_DIM,
depth=depth)
elif architecture == 'vgg16':
from mmdnn.conversion.examples.paddle.models import vgg
out = vgg.vgg16(image, class_dim=CLASS_DIM)
else:
print("Not support for {} yet.", architecture)
return None
architecture_file = path + architecture + '.bin'
paddle_extractor.dump_v2_config(out, architecture_file, True)
weight_file = download_file(
cls.architecture_map[architecture]['params'],
directory=path,
local_fname=architecture + '.tar.gz')
if not weight_file:
return None
print("MXNet Model {} saved as [{}] and [{}].".format(
architecture, architecture_file, weight_file))
return (architecture_file, weight_file)
else:
return None
def download(cls, architecture, path="./"):
if architecture in cls.thirdparty_map:
weight_file = download_file(cls.thirdparty_map[architecture],
directory=path)
return weight_file
elif cls.sanity_check(architecture):
output_filename = path + 'imagenet_{}.h5'.format(architecture)
if os.path.exists(output_filename) == False:
model = cls.architecture_map[architecture]()
model.save(output_filename)
print("Keras model {} is saved in [{}]".format(
architecture, output_filename))
K.clear_session()
del model
return output_filename
else:
print("File [{}] existed, skip download.".format(
output_filename))
return output_filename
else:
return None
def _main():
parser = argparse.ArgumentParser()
parser.add_argument('-n',
'--network',
type=_text_type,
help='Model Type',
required=True,
choices=MODEL_URL.keys())
parser.add_argument('-i',
'--image',
default=None,
type=_text_type,
help='Test Image Path')
parser.add_argument('-o',
'--output_dir',
default='./',
type=_text_type,
help='Paddlepaddle parameters file name')
args = parser.parse_args()
fn = download_file(MODEL_URL[args.network],
local_fname=architecture + '.tar.gz',
directory=args.output_dir)
if not fn:
return -1
DATA_DIM = 3 * IMG_SIZE * IMG_SIZE # Use 3 * 331 * 331 or 3 * 299 * 299 for Inception-ResNet-v2.
CLASS_DIM = CLASS_DIMS[args.network]
# refer to https://github.com/PaddlePaddle/Paddle/blob/develop/python/paddle/v2/tests/test_rnn_layer.py#L35
reset_parser()
# refer to https://github.com/PaddlePaddle/Paddle/issues/7403
paddle.init(use_gpu=False, trainer_count=1)
image = paddle.layer.data(name="image",
type=paddle.data_type.dense_vector(DATA_DIM))
if 'resnet' in architecture:
from mmdnn.conversion.examples.paddle.models import resnet
depth = int(architecture.strip('resnet'))
out = resnet.resnet_imagenet(image, class_dim=CLASS_DIM, depth=depth)
elif architecture == 'vgg16':
from mmdnn.conversion.examples.paddle.models import vgg
out = vgg.vgg16(image, class_dim=CLASS_DIM)
else:
print("Not support for {} yet.", architecture)
return None
dump_v2_config(out, args.output_dir + architecture + '.bin')
print("Model {} is saved as {} and {}.".format(
args.network, args.output_dir + architecture + '.bin', fn))
if args.image:
import numpy as np
from mmdnn.conversion.examples.imagenet_test import TestKit
func = TestKit.preprocess_func['paddle'][args.network]
img = func(args.image)
img = np.transpose(img, (2, 0, 1))
test_data = [(img.flatten(), )]
with gzip.open(parameters_file, 'r') as f:
parameters = paddle.parameters.Parameters.from_tar(f)
predict = paddle.infer(output_layer=out,
parameters=parameters,
input=test_data)
predict = np.squeeze(predict)
top_indices = predict.argsort()[-5:][::-1]
result = [(i, predict[i]) for i in top_indices]
print(result)
print(np.sum(result))
return 0
def _main():
parser = argparse.ArgumentParser()
parser.add_argument('-n',
'--network',
type=_text_type,
help='Model Type',
required=True,
choices=_default_model_info.keys())
parser.add_argument('-i',
'--image',
default=None,
type=_text_type,
help='Test Image Path')
parser.add_argument('-o',
'--output_dir',
default='./',
type=_text_type,
help='Tensorflow Checkpoint file name')
args = parser.parse_args()
if not download_file(_default_model_info[args.network]['symbol'],
directory=args.output_dir):
return -1
if not download_file(_default_model_info[args.network]['params'],
directory=args.output_dir):
return -1
print("Model {} saved.".format(args.network))
file_name = _default_model_info[args.network]['params'].split('/')[-1]
prefix, epoch_num = file_name[:-7].rsplit('-', 1)
sym, arg_params, aux_params = mx.model.load_checkpoint(
args.output_dir + prefix, int(epoch_num))
model = mx.mod.Module(symbol=sym)
model.bind(for_training=False,
data_shapes=[
('data',
(1, 3, _default_model_info[args.network]['image_size'],
_default_model_info[args.network]['image_size']))
])
model.set_params(arg_params,
aux_params,
allow_missing=True,
allow_extra=True)
if args.image:
import numpy as np
# need to be updated
network = _search_preprocess_key(args.network)
func = TestKit.preprocess_func['mxnet'][network]
img = func(args.image)
img = np.swapaxes(img, 0, 2)
img = np.swapaxes(img, 1, 2)
img = np.expand_dims(img, axis=0)
model.forward(Batch([mx.nd.array(img)]))
predict = model.get_outputs()[0].asnumpy()
predict = np.squeeze(predict)
top_indices = predict.argsort()[-5:][::-1]
result = [(i, predict[i]) for i in top_indices]
print(result)
return 0
