else:
to_test_net = Net.Lstm(model_file=conf['model_path'],
framework="keras")
else: # data_type == "Frames_dataset
sample_type = conf['data_path'].split('/')[-1]
data_type = data_type + "_" + sample_type
samples_dir = conf['data_path'].split('/')[5]
dim = (int(samples_dir.split('_')[-2]),
int(samples_dir.split('_')[-1]))
if sample_type == "raw_samples":
if net_type == "NOREC":
print('Puting the test data into the right shape...')
parameters, testX, testY = frame_utils.read_frame_data(
conf['data_path'], sample_type)
to_test_net = Net.Convolution2D(model_file=conf['model_path'],
framework="keras")
else:
print('Puting the test data into the right shape...')
parameters, testX, testY = frame_utils.read_frame_data(
conf['data_path'], sample_type, True)
to_test_net = Net.ConvolutionLstm(
model_file=conf['model_path'], framework="keras")
else:
parameters, testX, testY = frame_utils.read_frame_data(
conf['data_path'], sample_type)
if net_type == "NOREC":
print('Puting the test data into the right shape...')
to_test_net = Net.Mlp(model_file=conf['model_path'],
framework="tensorflow")
else:
print('Puting the test data into the right shape...')
in_dim = [images_per_sample, dim[0], dim[1], 1]
else:
in_dim = [images_per_sample, dim[0], dim[1]]
else:
_, trainX, trainY = frame_utils.read_frame_data(data_dir + 'train/', 'raw_samples', channels)
_, valX, valY = frame_utils.read_frame_data(data_dir + 'val/', 'raw_samples', channels)
train_data = [trainX, trainY]
val_data = [valX, valY]
in_dim = trainX.shape[1:]
out_dim = np.prod(in_dim[1:])
# Model settings
if net_type == "NoRec":
to_train_net = Net.Convolution2D(activation=activation, loss=loss, dropout=dropout,
drop_percentage=drop_percentage, input_shape=in_dim,
output_shape=out_dim, complexity=complexity, framework="keras")
else:
to_train_net = Net.ConvolutionLstm(activation=activation, loss=loss, dropout=dropout,
drop_percentage=drop_percentage, input_shape=in_dim,
output_shape=out_dim, complexity=complexity, framework="keras")
else:
print("Modeled images")
loss = conf['modeled_frame_loss']
activation = conf['modeled_activation']
dim = (int(samples_dir.split('_')[-2]), int(samples_dir.split('_')[-1]))
filename = root + "_Modeled/" + complexity
_, trainX, trainY = frame_utils.read_frame_data(data_dir + 'train/', 'modeled_samples')
_, valX, valY = frame_utils.read_frame_data(data_dir + 'val/', 'modeled_samples')
