def __init__ (self) : args = config.parse_arguments() self.roi_model=args.roi_model self.main_model=args.main_model self.load_weights_roi=args.load_weights_roi self.load_weights_main=args.load_weights_main self.store_model=args.store_txt self.batch_size=args.batch_size
def __init__(self): args = config.parse_arguments() self.store_model_path = args.store_txt self.classes = args.classes self.loss_main = args.loss_main self.loss_roi = args.loss_roi self.width = args.width self.height = args.height
def __init__(self, init1, init2): args = config.parse_arguments() self.roi_activation = args.roi_activation self.init_f = init1 self.init_b = init2 self.loss_pre = args.loss_pre self.roi_shape = args.roi_shape_roi self.roi_optimizer = args.pre_optimizer self.get_history = False self.verbose = 1 self.units_ed = 64 self.pretrain_window = args.pretrain_window self.epoch_pre = args.epochs_pre self.monitor = args.monitor_callbacks self.store_model = args.store_txt self.weight_name = 'first'
def __init__(self, X, Y): self.mask = Y self.images = X args = config.parse_arguments() self.max = args.max_loops self.featurewise_center = args.featurewise_center self.samplewise_center = args.samplewise_center self.featurewise_std_normalization = args.featurewise_std_normalization self.samplewise_std_normalization = args.samplewise_std_normalization self.zca_whitening = args.zca_whitening self.rotation_range = args.rotation_range self.width_shift_range = args.width_shift_range self.height_shift_range = args.height_shift_range self.horizontal_flip = args.horizontal_flip self.vertical_flip = args.vertical_flip self.data_augm = args.data_augm_classic self.alpha = args.alpha self.sigma = args.sigma self.normilize = args.normalize self.shuffle = args.shuffle self.batch_size = args.batch_size self.index = np.arange(len(self.images)) self.noise = args.noise self.random_apply_in_batch = args.random_apply_in_batch if self.data_augm == 'True': self.datagen = ImageDataGenerator( featurewise_center=self. featurewise_center, # set input mean to 0 over the dataset samplewise_center=self. samplewise_center, # set each sample mean o 0 featurewise_std_normalization=self. featurewise_std_normalization, # divide inputs by std of the dataset samplewise_std_normalization=self. samplewise_std_normalization, # divide each input by its std zca_whitening=self.zca_whitening, # apply ZCA whitening rotation_range=self. rotation_range, # randomly rotate images in the range (degrees, 0 to 180) width_shift_range=self. width_shift_range, # randomly shift images horizontally (fraction of total width) height_shift_range=self. height_shift_range, # randomly shift images vertically (fraction of total height) horizontal_flip=self.horizontal_flip, # randomly flip images vertical_flip=self.vertical_flip, zca_epsilon=1e-6)
def __init__(self, init1, init2, height, channels, classes, width): args = config.parse_arguments() self.height = height self.channels = channels self.classes = classes self.main_activation = args.main_activation self.init_w = init1 self.init_b = init2 self.features = args.features self.depth = args.depth self.padding = args.padding self.batchnorm = args.batchnorm self.dropout = args.dropout self.width = width self.temperature = 1.0 self.max_norm_const = args.max_norm_const self.max_norm_value = args.max_norm_value self.im_length = args.height
def __init__(self, rmn, mmn, data_return='off', save_mode='off'): args = config.parse_arguments() self.roi_model_name = rmn self.main_model_name = mmn self.data_return = data_return self.save_mode = save_mode self.STORE_PATH = args.store_data_test self.STORE_PATH1 = self.STORE_PATH + '/ROI/train/' self.STORE_PATH2 = self.STORE_PATH + '/ROI/test/' self.STORE_PATH_main1 = self.STORE_PATH + '/MAIN/train/' self.STORE_PATH_main2 = self.STORE_PATH + '/MAIN/test/' # if the paths does not excist create automatic # if not os.path.exists(self.STORE_PATH1): # os.makedirs(self.STORE_PATH1) # if not os.path.exists(self.STORE_PATH2): # os.makedirs(self.STORE_PATH2) # if not os.path.exists(self.STORE_PATH_main1): # os.makedirs(self.STORE_PATH_main1) # if not os.path.exists(self.STORE_PATH_main2): # os.makedirs(self.STORE_PATH_main2) self.data_extention = args.data_extention self.counter_extention = args.counter_extention self.restore_from_jpg_tif = args.restore_image self.label_classes = args.label_classes if self.label_classes == 'both': self.classes = 2 if self.label_classes == 'three': self.classes = 3 if self.label_classes == 'four': self.classes = 4 if self.label_classes == 'first': self.classes = 1 if self.label_classes == 'second': self.classes = 1 if self.label_classes == 'third': self.classes = 1 if self.label_classes == 'fourth': self.classes = 1
def __init__(self, X, Y, case): args = config.parse_arguments() self.case = case self.gan_train_directory = args.gan_train_directory self.data_augm = args.data_augm self.batch_size = args.batch_size self.batch_size_test = args.batch_size_test self.num_cores = args.num_cores self.validation_split = args.validation_split self.validation = args.validation self.shuffle = args.shuffle self.normalize_image = args.normalize self.gancheckpoint = 'checkpoint' self.gan_synthetic = args.gan_synthetic self.num_synthetic_images = args.num_synthetic_images self.batch_size = args.batch_size self.X = X self.Y = Y self.fourier = args.fft_convert_data self.STORE_PATH = args.store_data_test self.STORE_PATH10 = self.STORE_PATH + '/ROI/train/' self.STORE_PATH1 = self.STORE_PATH + '/FFT/train/' self.data_extention = "jpeg" boolt = 'False' self.main_model = args.main_model if (args.main_model == 'rgu_net' and self.case == 'main'): A, nodes_coordinates = grid_graph(args.height) coarsening_levels = args.depth u_shape, u_rows, u_cols, u_val, perm = coarsen_mnist( A, coarsening_levels, nodes_coordinates) self.X = convert_train_data(X, perm, args.height) print( 'the modified shape of X input because of RGMM structure is: ') print(self.X.shape) if boolt == 'True': self.threshold_bin = 0.8 else: self.threshold_bin = 0.05
import sys import cv2 import matplotlib.pyplot as plt from auto_segm import run_model, config from keras.losses import mean_squared_error import numpy as np import itertools from keras.models import Sequential from keras.layers import Dense from keras.utils import to_categorical from keras.wrappers.scikit_learn import KerasClassifier class vision_outputs: def __init__(self, case, history=None): args = config.parse_arguments() self.h = history self.STORE_TXT = args.store_txt self.store_results = args.store_results self.validation = args.validation self.metrics = args.metrics self.metrics1 = args.metrics1 self.metrics2 = args.metrics2 self.case = case def plot_loss(self): ''' plot the loss results ''' print(self.h.history.keys()) if (self.validation == "on"):
def __init__(self, path_case): args = config.parse_arguments() self.model_name = args.main_model self.ram = args.ram self.cross_validation_number = args.crossval_cycle self.ngpu = args.ngpu self.metrics = args.metrics self.metrics1 = args.metrics1 self.metrics2 = args.metrics2 self.batch_size = args.batch_size self.batch_size_test = args.batch_size_test self.epochs_roi = args.epochs_roi self.epochs_main = args.epochs_main self.num_cores = args.num_cores self.path_case = path_case self.validation_split = args.validation_split self.validation = args.validation self.shuffle = args.shuffle self.optimizer = args.pre_optimizer self.weights = args.loss_weights self.normalize_image = args.normalize self.roi_shape = args.roi_shape_roi self.store_model = args.store_txt self.weight_name = 'first' self.early_stopping = args.early_stop self.monitor = args.monitor_callbacks self.mode = args.mode_convert self.label_classes = args.label_classes self.exponential_decay = 'False' self.lrate = args.learning_rate self.store_model_path = args.store_txt self.fourier = args.fft_convert_data if (args.decay == 666): self.exponential_decay = 'True' args.decay = 0 optimizer_args_roi = { 'lr': args.roi_learning_rate, 'momentum': args.roi_momentum, 'decay': args.roi_decay, 'seed': args.roi_seed } optimizer_args = { 'lr': args.learning_rate, 'momentum': args.momentum, 'decay': args.decay, 'seed': args.seed } if self.path_case == 'roi': for k in list(optimizer_args_roi): if optimizer_args_roi[k] is None: del optimizer_args_roi[k] optimizer = self.pass_optimizer(args.roi_optimizer, optimizer_args_roi) self.optimizer = optimizer self.epochs = self.epochs_roi if self.path_case == 'main': for k in list(optimizer_args): if optimizer_args[k] is None: del optimizer_args[k] optimizer = self.pass_optimizer(args.m_optimizer, optimizer_args) self.optimizer = optimizer self.epochs = self.epochs_main
def __init__ (self,analysis,path_case) : """ Initializare of the config file """ args = config.parse_arguments() self.path_case=path_case self.restore_from_jpg_tif=args.restore_image self.rotated="False" if self.path_case=='roi': self.image_shape=args.image_shape_roi self.original_image_shape=args.original_image_shape_roi self.roi_shape=args.roi_shape_roi self.data_path=args.datapath self.data_path2=args.datapath self.STORE_TXT=args.store_txt self.counter_path='/contour/' self.data_extention = args.data_extention_roi self.counter_extention = args.counter_extention_roi self.PATH_IMAGES='/image' self.PATH_IMAGES2='/image' if self.path_case=='main': if self.restore_from_jpg_tif=='off': self.image_shape=args.image_shape self.original_image_shape=args.original_image_shape self.roi_shape=args.roi_shape self.data_path=args.datapath self.data_path2=args.datapath self.STORE_TXT=args.store_txt self.counter_path='/contour/' self.data_extention = args.data_extention_roi self.counter_extention = args.counter_extention_roi self.PATH_IMAGES='/image' self.PATH_IMAGES2='/image' else: self.image_shape=args.image_shape self.original_image_shape=args.original_image_shape self.roi_shape=args.roi_shape self.data_path=args.store_data_test self.data_path2=args.datapath self.STORE_TXT=args.store_txt self.counter_path='/contour_main/' self.data_extention = args.data_extention self.counter_extention = args.counter_extention self.PATH_IMAGES='/ROI/train' self.PATH_IMAGES2='/ROI/test' if self.path_case=='pre': self.image_shape=args.image_shape_pre self.original_image_shape=args.original_image_shape_pre self.roi_shape=args.roi_shape_pre self.data_path=args.datapath self.data_path2=args.datapath self.STORE_TXT=args.store_txt self.counter_path='/contour/' self.data_extention = args.data_extention_pre self.counter_extention = args.counter_extention_pre self.PATH_IMAGES='/image' self.PATH_IMAGES2='/image' self.pretrain_window=args.pretrain_window # seperate the train of ROI with the train set for the u_net. Thus take the epi and endo seperate in u_net and train the ROI detection in both if (analysis=='train' or analysis=='test') : self.n_set_pre=analysis self.n_set=analysis else: self.n_set_pre='train_prediction' self.n_set='train' self.patient_list=args.patient_list self.store_contour=args.store_data_test self.image_part = np.zeros([self.original_image_shape,self.original_image_shape]) self.shuffle=args.shuffle self.num_preprocess_threads=args.num_cores self.batch_size=args.batch_size self.patient_store_style=args.patient_store_style self.STORE_PATH=args.store_data_test #if MICCAI_2009: self.label_case1=args.label_case_extension_1 self.label_case2=args.label_case_extension_2 self.label_classes=args.label_classes if self.label_classes=='both': self.classes=2 elif self.label_classes=='three': self.classes=3 elif self.label_classes=='four': self.classes=4 elif self.label_classes=='first': self.classes=1 elif self.label_classes=='second': self.classes=1 elif self.label_classes=='third': self.classes=1 elif self.label_classes=='fourth': self.classes=1 else: print("No accebtable label_classes!! ")