if not os.path.exists(args.output):
os.makedirs(args.output)
print('-------------')
print('BATCH : {}'.format(args.batch))
print('EPOCH : {}'.format(args.epoch))
print('ALPA : {}'.format(args.alpha))
print('DROPOUT : {}'.format(args.dropout))
print('Load Weights?: {}'.format(args.weights))
print('Dataset : {}'.format(args.dataset))
print('OUTPUT : {}'.format(args.output))
print('-------------')
# TODO: abstract method to normalize speed.
df_train, df_val = getDataFromThunderhill(args.dataset, split=True, randomize=True, balance=True)
print('TRAIN:', len(df_train))
print('VALIDATION:', len(df_val))
print(df_train[['speed', 'throttle', 'brake', 'accel']].describe())
model = NvidiaModel(args.dropout)
print(model.summary())
# Saves the model...
with open(os.path.join(args.output, 'model.json'), 'w') as f:
f.write(model.to_json())
try:
if args.weights:
print('Loading weights from file ...')
model.load_weights(args.weights)
if not os.path.exists(args.output):
os.makedirs(args.output)
print('-------------')
print('BATCH : {}'.format(args.batch))
print('EPOCH : {}'.format(args.epoch))
print('ALPA : {}'.format(args.alpha))
print('DROPOUT : {}'.format(args.dropout))
print('Load Weights?: {}'.format(args.weights))
print('Dataset : {}'.format(args.dataset))
print('OUTPUT : {}'.format(args.output))
print('-------------')
# TODO: abstract method to normalize speed.
df_train, df_val = getDataFromThunderhill(args.dataset,
args.output,
balance=False)
print('TRAIN:', len(df_train))
print('VALIDATION:', len(df_val))
print(df_train.describe())
model = NvidiaModel(args.dropout)
print(model.summary())
# Saves the model...
with open(os.path.join(args.output, 'model.json'), 'w') as f:
f.write(model.to_json())
try:
if args.weights:
print('Loading weights from file ...')