path_to_img = args.img
path_to_model = args.m
path_to_csv_output = args.o
path_to_figures = args.f
# initiate the model
if args.t == "du":
model = DenseNet.DenseUnet_v2(weights=None,
input_shape=[128, 128, 3],
loss="compound")
model.load_weights(path_to_model)
elif args.t == "u":
model = cnn.Unet()
model.initiate(128, 128, 3)
model.load(path_to_model)
else:
print(
"specify which type of model to use (u for unet and du for dense unet)"
)
# load images
images = os.listdir(path_to_img)
# initiate an empty dataframe
cnn_prediction = pd.DataFrame(columns=['x', 'y', 'uniqueframe'])
# plot every Nth frame
N = int(args.n)
def initialize(path_to_model):
import cnn_model as cnn
model = cnn.Unet()
model.initiate()
model.load(path_to_model)
return model
#!/usr/bin/env python3 # -*- coding: utf-8 -*- """ Created on Sat Apr 28 11:53:49 2018 @author: luke """ import numpy as np import cnn_model as cnn unet = cnn.Unet() unet.initiate() unet.plot(r'/home/luke/Videos/ultrasound/src/model.png') x = np.load(r'/home/luke/Videos/ultrasound/data/training.npy') y10 = np.load(r'/home/luke/Videos/ultrasound/data/y_10.npy') path_to_model_y10 = r'/home/luke/Videos/ultrasound/data/model_y10.hdf5' path_to_csv_y10 = r'/home/luke/Videos/ultrasound/data/log_y10.csv' unet = cnn.Unet() unet.initiate() unet.train(x, y10, path_to_model_y10, path_to_csv_y10)