def _generate_predictions(model, data_generator, batch_size, nb_samples,
vol_params):
# whole volumes
if vol_params is not None:
for _ in range(nb_samples): # assumes nr volume
vols = nrn_utils.predict_volumes(model, data_generator, batch_size,
vol_params["patch_size"],
vol_params["patch_stride"],
vol_params["grid_size"])
vol_true, vol_pred = vols[0], vols[1]
yield (vol_true, vol_pred)
# just one batch
else:
for _ in range(nb_samples): # assumes nr batches
vol_pred, vol_true = nrn_utils.next_label(model, data_generator)
yield (vol_true, vol_pred)
def _generate_predictions(model, data_generator, batch_size, nb_samples, vol_params):
# whole volumes
if vol_params is not None:
for _ in range(nb_samples): # assumes nr volume
vols = nrn_utils.predict_volumes(model,
data_generator,
batch_size,
vol_params["patch_size"],
vol_params["patch_stride"],
vol_params["grid_size"])
vol_true, vol_pred = vols[0], vols[1]
yield (vol_true, vol_pred)
# just one batch
else:
for _ in range(nb_samples): # assumes nr batches
vol_pred, vol_true = nrn_utils.next_label(model, data_generator)
yield (vol_true, vol_pred)