train_chase.py

import os

import numpy as np
import cv2
from sklearn.model_selection import train_test_split
from keras.callbacks import TensorBoard, ModelCheckpoint
from  util import *
np.random.seed(42)
import scipy.misc as mc
data_location = ''
training_images_loc = data_location + 'Chase/train/image/'
training_label_loc = data_location + 'Chase/train/label/'
train_files = os.listdir(training_images_loc)
train_data = []
train_label = []
validate_data = []
validate_label = []
desired_size=1008
for i in train_files:
    im = mc.imread(training_images_loc + i)
    label = mc.imread(training_label_loc + "Image_" +i.split('_')[1].split(".")[0] +"_1stHO.png" )
    old_size = im.shape[:2]  # old_size is in (height, width) format
    delta_w = desired_size - old_size[1]
    delta_h = desired_size - old_size[0]

    top, bottom = delta_h // 2, delta_h - (delta_h // 2)
    left, right = delta_w // 2, delta_w - (delta_w // 2)

    color = [0, 0, 0]
    color2 = [0]
    new_im = cv2.copyMakeBorder(im, top, bottom, left, right, cv2.BORDER_CONSTANT,
                                value=color)

    new_label = cv2.copyMakeBorder(label, top, bottom, left, right, cv2.BORDER_CONSTANT,
                                   value=color2)

    train_data.append(cv2.resize(new_im, (desired_size, desired_size)))

    temp = cv2.resize(new_label,
                      (desired_size, desired_size))
    _, temp = cv2.threshold(temp, 127, 255, cv2.THRESH_BINARY)
    train_label.append(temp)

train_data = np.array(train_data)
train_label = np.array(train_label)



x_train = train_data.astype('float32') / 255.
y_train = train_label.astype('float32') / 255.


x_rotated, y_rotated, x_flipped, y_flipped = img_augmentation(x_train, y_train)

x_train= np.concatenate([x_train, x_rotated,x_flipped])
y_train = np.concatenate([y_train, y_rotated,y_flipped])



x_train, x_validate, y_train, y_validate = train_test_split(x_train, y_train, test_size = 0.10, random_state = 101)

x_train = np.reshape(x_train, (len(x_train), desired_size, desired_size, 3))  # adapt this if using `channels_first` image data format
y_train = np.reshape(y_train, (len(y_train), desired_size, desired_size, 1))  # adapt this if using `channels_first` im

x_validate = np.reshape(x_validate, (len(x_validate), desired_size, desired_size, 3))  # adapt this if using `channels_first` image data format
y_validate = np.reshape(y_validate, (len(y_validate), desired_size, desired_size, 1))  # adapt this if using `channels_first` im


TensorBoard(log_dir='./autoencoder', histogram_freq=0,
            write_graph=True, write_images=True)

from  CARUNet import *
model=CARUNet(input_size=(desired_size,desired_size,3),start_neurons=16,keep_prob=0.85,lr=1e-3)
weight="Chase/Model/CARUNet.h5"

restore=False

if restore and os.path.isfile(weight):
    model.load_weights(weight)

model_checkpoint = ModelCheckpoint(weight, monitor='val_accuracy', verbose=1, save_best_only=False)

history=model.fit(x_train, y_train,
                epochs=50,
                batch_size=1,
                # validation_split=0.1,
                validation_data=(x_validate, y_validate),
                shuffle=True,
                callbacks= [TensorBoard(log_dir='./autoencoder'), model_checkpoint])