default_transfo_params = model.default_transfo_params
no_transfo_params = model.no_transfo_params
if hasattr(model, 'paired_transfos'):
paired_transfos = model.paired_transfos
else:
paired_transfos = False
data_loader = DataLoader(
images_train_0=images_train_0,
labels_train_0=labels_train_0,
images_train_1=images_train_1,
labels_train_1=labels_train_1,
images_train_eval=images_train_eval,
labels_train_eval=labels_train_eval,
images_valid_eval=images_valid_eval,
labels_valid_eval=labels_valid_eval,
p_x=model.output_size,
p_y=model.output_size,
num_channels=model.num_channels,
prefix_train=prefix_train,
prefix_test=prefix_test,
default_transfo_params=default_transfo_params,
no_transfo_params=no_transfo_params,
)
print "Estimating parameters ..."
start = time.time()
if hasattr(model, 'pixel_based_norm'):
pixel_based_norm = model.pixel_based_norm
else:
test_ids.append(test_id)
return sorted(set(test_ids))
img_ids = get_img_ids(img_dir)
if len(img_ids) == 0:
raise ValueError('No img ids!\n')
print("\n\nDoing prediction on %s set.\n" % dataset)
print("\n\t%i test ids.\n" % len(img_ids))
# Create dataloader with the test ids.
from generators import DataLoader
data_loader = DataLoader() # model_data['data_loader']
new_dataloader_params = model_data['data_loader_params']
new_dataloader_params.update({'images_test': img_ids})
data_loader.set_params(new_dataloader_params)
if 'paired_transfos' in model_data:
paired_transfos = model_data['paired_transfos']
else:
paired_transfos = False
print("\tChunk size: %i.\n" % chunk_size)
num_chunks = int(np.ceil((2 * len(img_ids)) / float(chunk_size)))
if 'data_loader_no_transfos' in model_data:
no_transfo_params = model_data['data_loader_no_transfos']
print('Dumped new model')
output = nn.layers.get_output(l_out, deterministic=True)
input_ndims = [len(nn.layers.get_output_shape(l_in)) for l_in in l_ins]
xs_shared = [nn.utils.shared_empty(dim=ndim) for ndim in input_ndims]
idx = T.lscalar('idx')
givens = {}
for l_in, x_shared in zip(l_ins, xs_shared):
givens[l_in.input_var] = x_shared[idx * batch_size:(idx + 1) * batch_size]
compute_output = theano.function([idx],
output,
givens=givens,
on_unused_input='ignore')
data_loader = DataLoader()
new_dataloader_params = model_data['data_loader_params']
new_dataloader_params.update({'images_test': patient_ids})
new_dataloader_params.update({'labels_test': train_labels.level.values})
new_dataloader_params.update({'prefix_train': IMG_DIR})
data_loader.set_params(new_dataloader_params)
num_chunks = int(np.ceil((2 * len(patient_ids)) / float(chunk_size)))
def do_pred(img_ids):
test_gen = lambda: data_loader.create_fixed_gen(
img_ids,
chunk_size=chunk_size,
prefix_train=IMG_DIR,
prefix_test=IMG_DIR,
transfo_params=no_transfo_params,
