def test_load_default_model():
machine = MachineFactory()
model = machine.load_model()
assert isinstance(model, CNNClassificationModel)
assert isinstance(model, Resnet50CNNModel)
def test_load_model():
machine = MachineFactory()
model = machine.load_model(
"pyspec.machine.model.application.Resnet50CNNModel")
assert isinstance(model, CNNClassificationModel)
assert isinstance(model, Resnet50CNNModel)
def test_load_encoder():
machine = MachineFactory()
encoder = machine.load_encoder()
assert encoder.width == 500
assert encoder.height == 500
assert encoder.axis is False
assert encoder.intensity_max == 1000
assert encoder.min_mz == 0
assert encoder.max_mz == 2000
assert encoder.dpi == 72
def load_model():
global clean_dirty_model
global clean_dirty_encoder
global graph
share = S3Share(read_only=True)
share.retrieve("clean_dirty", force=True)
factory = MachineFactory()
# get model and encoder as configured
clean_dirty_model = factory.load_model()
clean_dirty_encoder = factory.load_encoder()
graph = tf.get_default_graph()
required=False,
type=str,
default=None)
parser.add_argument("--dataset",
help="path to your dataset",
required=True,
type=str)
parser.add_argument("--configuration",
help="which configuration file to use",
required=True,
type=str)
parser.add_argument("--gpu",
help="which gpu to use, by deault all will be utilized",
type=int,
default=-1)
args = parser.parse_args()
factory = MachineFactory(config_file=args.configuration)
model = factory.load_model(name=args.model)
if args.gpu < 0:
factory.train(args.dataset, model=model, gpus=get_gpu_count())
else:
os.environ['CUDA_VISIBLE_DEVICES'] = "{}".format(args.gpu)
factory.train(args.dataset, model=model, gpus=1)
# uploading dataset to cloud storage for reuse by other software
share = S3Share()
share.submit(args.dataset)
type=str)
parser.add_argument("--configuration",
help="which configuration file to use",
required=False,
type=str,
default="machine.ini")
parser.add_argument("--gpu", help="which gpu to use", type=int, default=-1)
parser.add_argument("--model",
help="model you would like to predict",
required=False,
type=str,
default=None)
args = parser.parse_args()
factory = MachineFactory(config_file=args.configuration)
model = factory.load_model(name=args.model)
def callback(file, classname, full_path: str):
if classname == 0:
classname = "clean"
else:
classname = "dirty"
print("{} is {}".format(file, classname))
os.makedirs("{}/{}".format("datasets/{}/sorted".format(args.dataset),
classname),
exist_ok=True)
copyfile(
def test_train_model():
machine = MachineFactory()
machine.train("datasets/clean_dirty", gpus=1)
def test_load_generator():
machine = MachineFactory()
generator = machine.load_generator()
assert isinstance(generator, LabelGenerator)