def __init__(self):
self.num_classes = 10
self.batch_size = 128
self.model_path = args.path_cifar10_vgg_model
create_dir_if_needed(self.model_path)
self.model_pef_path = args.path_cifar10_vgg_model_perf
create_dir_if_needed(self.model_pef_path)
self.epochs = 150
self.img_rows, self.img_cols, self.img_chns = 32, 32, 3
(self.x_train, self.y_train), (self.x_test,
self.y_test) = cifar10.load_data()
self.pre_reshape()
self.weight_decay = 0.0005
self.x_shape = [32, 32, 3]
self.model = self.get_model_archi()
#self.weights = self.get_model_weights()
print(self.x_train.shape)
print(self.x_test.shape)
def __init__(self):
self.num_classes = 10
self.batch_size = 32
self.model_path = args.path_cifar10_resnet_model
create_dir_if_needed(self.model_path)
self.model_pef_path = args.path_cifar10_resnet_model_perf
create_dir_if_needed(self.model_pef_path)
self.epochs = 200
self.img_rows, self.img_cols, self.img_chns = 32, 32, 3
(self.x_train, self.y_train), (self.x_test,
self.y_test) = cifar10.load_data()
self.pre_reshape()
self.weight_decay = 0.0005
self.x_shape = [32, 32, 3]
self.model = self.get_model_archi()
# self.weights = self.get_model_weights()
self.lr_reducer = ReduceLROnPlateau(factor=np.sqrt(0.1),
cooldown=0,
patience=5,
min_lr=0.5e-6)
self.early_stopper = EarlyStopping(min_delta=0.001, patience=10)
self.csv_logger = CSVLogger('resnet18_cifar10.csv')
print(self.x_train.shape)
print(self.x_test.shape)
def __init__(self):
self.num_classes = 10
self.batch_size = 128
self.model_path = args.path_cifar10_vgg_model
create_dir_if_needed(self.model_path)
self.model_pef_path = args.path_cifar10_vgg_model_perf
create_dir_if_needed(self.model_pef_path)
self.epochs = 200
self.img_rows, self.img_cols, self.img_chns = 32, 32, 3
(self.x_train, self.y_train), (self.x_test,
self.y_test) = cifar10.load_data()
self.weight_decay = 0.0005
self.x_shape = [32, 32, 3]
self.mean = 120.707
self.std = 64.15
self.model = self.get_model_archi()
#self.weights = self.get_model_weights()
#training parameters
self.lr = 0.1
self.lr_decay = 1e-6
self.lr_drop = 20
print(self.x_train.shape)
print(self.x_test.shape)
def __init__(self):
self.num_classes = 10
self.img_rows, self.img_cols, self.img_chns = 28, 28, 1
self.model = self.get_model_archi()
self.model_path = args.path_mnist_model
create_dir_if_needed(self.model_path)
self.model_pef_path = args.path_mnist_model_perf #not necessary!!
create_dir_if_needed(self.model_pef_path)
self.batch_size = 128
self.epochs = 3
(self.x_train, self.y_train), (self.x_test,
self.y_test) = mnist.load_data()
def __init__(self):
self.model_path = args.path_cifar10_2c2d_model
create_dir_if_needed(self.model_path)
self.model_pef_path = args.path_cifar10_2c2d_model_perf
create_dir_if_needed(self.model_pef_path)
self.num_classes = 10
self.batch_size = 128
self.epochs = 25
self.img_rows, self.img_cols, self.img_chns = 32, 32, 3
(self.x_train, self.y_train), (self.x_test,
self.y_test) = cifar10.load_data()
self.pre_reshape()
self.model = self.get_model_archi()
def __init__(self):
self.num_classes = 10
self.batch_size = 128
self.model_path = args.path_cifar10_2c1d_model
create_dir_if_needed(self.model_path)
self.model_pef_path = args.path_cifar10_2c1d_model_perf
create_dir_if_needed(self.model_pef_path)
self.epochs = 25
self.img_rows, self.img_cols, self.img_chns = 32, 32, 3
(self.x_train, self.y_train), (self.x_test,
self.y_test) = cifar10.load_data()
#shape: (50000, 32, 32, 3) (50000, 1) (10000, 32, 32, 3) (10000, 1)
self.mean = 120.707
self.std = 64.15
self.model = self.get_model_archi()
import logging
import keras
from keras.datasets import mnist, cifar10
from keras.models import Sequential, model_from_json, load_model
from keras.layers import Dense, Dropout, Flatten
from keras.layers import Conv2D, MaxPooling2D
from keras import backend as K
from my_funs import print_acc_per_class, create_dir_if_needed
from my_models import Cifar10_2c2d, Mnist_2c1d, Cifar10_vgg, Cifar10_resnet
from parameters import *
# create the needed directories
for i in [model_dir, log_dir, img_dir]:
create_dir_if_needed(i)
# global variables
# create log at terminal and disk at the same time
logging.basicConfig(
level=logging.INFO,
format="%(asctime)s [%(levelname)-5.5s] %(message)s",
handlers=[logging.FileHandler(args.log_file),
logging.StreamHandler()])
def create_dir_if_needed(dest_directory):
"""
Create directory if doesn't exist
:param dest_directory:
:return: True if everything went well
def set_saliency_path(self, path):
self.saliency_path = path
create_dir_if_needed(path)
def set_model_path(self, path):
self.model_path = path
create_dir_if_needed(path)