def train():
ROOT = './'
VGG16_WEIGHT_PATH = './vgg/vgg16_weights.npz'
DATASET_PATH = os.path.join(ROOT, 'VOC2012/')
CHECKPOINT_DIR = os.path.join(DATASET_PATH, 'saved_model')
IMAGE_SHAPE = (512, 512)
N_CLASSES = 21
N_EPOCHS = 100
BATCH_SIZE = 1
LEARNING_RATE = 1e-5
DECAY_RATE = 0.95
DECAY_EPOCH = 10
DROPOUT_RATE = 0.5
print('Starting end-to-end training FCN-8s')
session_config = tf.ConfigProto(gpu_options=tf.GPUOptions(
allow_growth=True))
session = tf.compat.v1.InteractiveSession(config=session_config)
session.as_default()
# ------------- Load VOC from TFRecord ---------------
dataset = VOCDataset()
dataset_train = dataset.load_dataset(DATASET_PATH,
BATCH_SIZE,
is_training=True)
dataset_val = dataset.load_dataset(DATASET_PATH,
BATCH_SIZE,
is_training=False)
# ------------- Build fcn model ------------
fcn = FCN(IMAGE_SHAPE, N_CLASSES, VGG16_WEIGHT_PATH)
fcn.build_from_vgg()
learning_rate_fn = learning_rate_with_exp_decay(BATCH_SIZE,
dataset.n_images['train'],
DECAY_EPOCH, DECAY_RATE,
LEARNING_RATE)
compile_model(fcn, learning_rate_fn)
fit_model(fcn, N_EPOCHS, BATCH_SIZE, dataset_train, dataset_val,
CHECKPOINT_DIR, DROPOUT_RATE)
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
tf.logging.set_verbosity(tf.logging.ERROR)
if __name__ == '__main__':
image_shape = (512, 512)
n_classes = 21
vgg16_weights_path = './vgg/vgg16_weights.npz'
model = FCN(image_shape, n_classes, vgg16_weights_path)
model.build_from_vgg()
root_path = './'
dataset_path = os.path.join(root_path, 'VOC2012/')
dataset = VOCDataset(augmentation_params=None)
dataset_val = dataset.load_dataset(dataset_path,
batch_size=8,
is_training=False)
iterator = tf.data.Iterator.from_structure(dataset_val.output_types,
dataset_val.output_shapes)
next_batch = iterator.get_next()
val_init_op = iterator.make_initializer(dataset_val)
session_config = tf.ConfigProto(gpu_options=tf.GPUOptions(
allow_growth=True))
with tf.Session(config=session_config) as session:
session.run(tf.global_variables_initializer())
session.run(val_init_op)
while True:
try: