import theano
import facial_recognition.util as util
from lasagne import layers
from nolearn.lasagne import NeuralNet
from facial_recognition.util import AdjustVariable
from facial_recognition.util import EarlyStopping
from facial_recognition.util import FlipBatchIterator
from facial_recognition.util import float32
try:
from lasagne.layers.cuda_convnet import Conv2DCCLayer as Conv2DLayer
from lasagne.layers.cuda_convnet import MaxPool2DCCLayer as MaxPool2DLayer
except ImportError:
Conv2DLayer = layers.Conv2DLayer
MaxPool2DLayer = layers.MaxPool2DLayer
def get_net():
return NeuralNet(
layers=[
('input', layers.InputLayer),
('conv1', Conv2DLayer),
('pool1', MaxPool2DLayer),
('dropout1', layers.DropoutLayer),
('conv2', Conv2DLayer),
('pool2', MaxPool2DLayer),
('dropout2', layers.DropoutLayer),
('conv3', Conv2DLayer),
('pool3', MaxPool2DLayer),
('dropout3', layers.DropoutLayer),
('hidden4', layers.DenseLayer),
('dropout4', layers.DropoutLayer),
('hidden5', layers.DenseLayer),
('output', layers.DenseLayer),
],
input_shape=(None, 1, 96, 96),
conv1_num_filters=32, conv1_filter_size=(3, 3), pool1_pool_size=(2, 2),
dropout1_p=0.1,
conv2_num_filters=64, conv2_filter_size=(2, 2), pool2_pool_size=(2, 2),
dropout2_p=0.2,
conv3_num_filters=128, conv3_filter_size=(2, 2), pool3_pool_size=(2, 2),
dropout3_p=0.3,
hidden4_num_units=1000,
dropout4_p=0.5,
hidden5_num_units=1000,
output_num_units=30, output_nonlinearity=None,
update_learning_rate=theano.shared(float32(0.03)),
update_momentum=theano.shared(float32(0.9)),
regression=True,
batch_iterator_train=FlipBatchIterator(batch_size=128),
on_epoch_finished=[
AdjustVariable('update_learning_rate', start=0.03, stop=0.0001),
AdjustVariable('update_momentum', start=0.9, stop=0.999),
EarlyStopping(patience=200),
],
max_epochs=3000,
verbose=1,
)
def train_network(net, save_name=''):
print("Loading data...")
X, y = util.load2d(util.FTRAIN)
print("Building network...")
print("Started training...")
net.fit(X, y)
print("Finished training...")
print("Saving network...")
util.pickle_network(save_name + ".pkl", net)
util.visualize_learning(net)
def load_and_visualize_network(file):
print("Loading data...")
X, y = util.load2d(util.FTEST)
print("Loading model...")
net = util.unpickle_network(file)
print("Finished training...")
# util.visualize_learning(net)
util.visualize_predictions(net)
net = get_net()
train_network(net, "net")
