MODEL_PATH = r"C:\NN\Experiments\DropletDetection\2020-01-29 07-55-29 - MediumNet_run2\checkpoint\best_epoch-00023_val-loss-0.01.hdf5"
DATA_FOLDER = r"D:\2805 RMED Pre-project Droplet Detection Deep Learning\_Cutout\All"
print_gpus()
# Load model / modify
def model_modifier(m):
m.layers[-1].activation = tf.keras.activations.linear # Replace softmax with linear. Somehow required.
model = load_model(MODEL_PATH)
model.summary()
# Load data
from tensorflow.keras.preprocessing.image import ImageDataGenerator
rescale_gen = ImageDataGenerator(rescale=1. / 255)
data_gen = rescale_gen.flow_from_directory(DATA_FOLDER, batch_size=1, class_mode='categorical', target_size=[224, 224])
image, label_one_hot = next(data_gen)
loss = lambda output: K.mean(output[:, label_one_hot.argmax()])
X = image
# Saliency
saliency = Saliency(model, model_modifier)
saliency_map = saliency(loss, X, smooth_samples=20)
saliency_map = normalize(saliency_map)
# plt.figure()
# plt.imshow((X[0]*255).astype(np.uint8))
# plt.figure()
# plt.imshow(saliency_map[0])
plt.figure()
classifier.add(Dense(units=1, activation='sigmoid'))
classifier.compile(optimizer='adam',
loss='binary_crossentropy',
metrics=['accuracy'])
from tensorflow.keras.preprocessing.image import ImageDataGenerator
train_datagen = ImageDataGenerator(rescale=1. / 255,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True)
test_datagen = ImageDataGenerator(rescale=1. / 255)
training_set = train_datagen.flow_from_directory('dataset/training_set',
target_size=(64, 64),
batch_size=32,
class_mode='binary')
test_set = test_datagen.flow_from_directory('dataset/test_set',
target_size=(64, 64),
batch_size=32,
class_mode='binary')
classifier.fit_generator(training_set,
steps_per_epoch=8000 / 32,
epochs=50,
validation_data=test_set,
validation_steps=2000 / 32)
# CREATING THE NEURAL NETWORK AND EMOTION DETECTOR
# 0. Importing the necessary libraries
import cv2
import numpy as np
import tensorflow as tf
from tensorflow.keras.preprocessing import image
from tensorflow.keras.preprocessing.image import ImageDataGenerator
# 1. Preprocessing the images/data
train_datagen = ImageDataGenerator(rescale=1. / 255)
train_set = train_datagen.flow_from_directory('your_folder_path/data/training',
target_size=(48, 48),
batch_size=32,
color_mode='grayscale',
class_mode='categorical')
test_datagen = ImageDataGenerator(rescale=1. / 255)
test_set = test_datagen.flow_from_directory('your_folder_path/data/testing',
target_size=(48, 48),
batch_size=32,
color_mode='grayscale',
class_mode='categorical')
# 2. Creating the model
cnn = tf.keras.models.Sequential()
#layer1
cnn.add(
tf.keras.layers.Conv2D(filters=32,
kernel_size=3,
rankdir="TB",
expand_nested=False,
dpi=96,
)
train = ImageDataGenerator(rescale=1.0 / 255,
rotation_range=40,
width_shift_range=0.2,
height_shift_range=0.2,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True,
fill_mode='nearest')
train_generator = train.flow_from_directory('./archive/train',
batch_size=64,
target_size=(48, 48),
class_mode='categorical')
validation = ImageDataGenerator(rescale=1.0 / 255)
validation_generator = validation.flow_from_directory('./archive/test',
batch_size=64,
target_size=(48, 48),
class_mode='categorical')
history = model.fit_generator(train_generator,
epochs=50,
steps_per_epoch=24176 // 64,
verbose=1,
validation_data=validation_generator,
validation_steps=6043 // 64,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True,
#vertical_flip=True,
width_shift_range=0.1,
height_shift_range=0.1,
#channel_shift_range=0.1,
#brightness_range=[0.1, 10]
)
# Note that the validation data should not be augmented!
test_datagen = ImageDataGenerator(rescale=1. / 255)
train_generator = train_datagen.flow_from_directory(train_dir,
target_size=(IMG_WIDTH,
IMG_HEIGHT),
batch_size=BATCH_SIZE,
class_mode='binary')
validation_generator = valid_datagen.flow_from_directory(
validation_dir,
target_size=(IMG_WIDTH, IMG_HEIGHT),
batch_size=BATCH_SIZE,
class_mode='binary')
# Horovod: write logs on worker 0.
verbose = 1 if hvd.rank() == 0 else 0
# Horovod: save checkpoints only on worker 0 to prevent other workers from corrupting them.
if hvd.rank() == 0:
callbacks.append(
