def get_models():
global _models
if _models is None:
model_path = app.config["TRAINED_MODEL_PATH"]
checkpoints = sorted(os.listdir(model_path))
_models = [load_from_convnet(os.path.join(model_path, c)) for c in checkpoints]
return _models
def get_models():
global _models
if _models is None:
model_path = app.config["TRAINED_MODEL_PATH"]
checkpoints = sorted(os.listdir(model_path))
_models = [
load_from_convnet(os.path.join(model_path, c)) for c in checkpoints
]
return _models
def _process_model(x):
# Import is here so that we don't need to have the convnet scripts
# on PYTHONPATH in order to interact with already-built databases.
from deepviz_webui.utils.decaf import load_from_convnet
(timestep, model_filename) = x
(directory, image_data, image_classes, num_classes) = _shared_data
_log.info("Processing model for timestep %i" % timestep)
model = load_from_convnet(model_filename)
stats = ModelStats.create(model, image_data, image_classes, num_classes)
stats.save(os.path.join(directory, str(timestep)))
def _process_model(x):
# Import is here so that we don't need to have the convnet scripts
# on PYTHONPATH in order to interact with already-built databases.
from deepviz_webui.utils.decaf import load_from_convnet
(timestep, model_filename) = x
(directory, image_data, image_classes, num_classes) = _shared_data
_log.info("Processing model for timestep %i" % timestep)
model = load_from_convnet(model_filename)
stats = ModelStats.create(model, image_data, image_classes, num_classes)
stats.save(os.path.join(directory, str(timestep)))
