def test_alexnet():
from keras.models import load_model
import numpy as np
import time
test_dir= '/home/pdd/pdwork/CV_BiShe/picture/trainTest/'
test, test_label = get_files(test_dir, 224, 224)
TP = 0
TN = 0
FP = 0
FN = 0
model = load_model('resnet50.h5')
time_start=time.time()
test_result = model.predict(test)
time_end=time.time()
time_spent = time_end-time_start
print(time_spent)
#calc precision and recall
for i in range(0,len(test_label)):
if np.argmax(test_result[i]) == 1:
if np.argmax(test_label[i]) == 1:
TP += 1
else:
FP += 1
else:
if np.argmax(test_label[i]) == 1:
FN += 1
else:
TN += 1
precision = TP/(FP+TP)
recall = TP/(FN+TP)
print('precision', precision)
print('recall', recall)
return time_spent, precision, recall
FN = 0
model = load_model('vgg16.h5')
test_result = model.predict(test)
time_end = time.time()
time_spent = time_end - time_start
print(time_spent)
#calc precision and recall
for i in range(0, len(test_label)):
if np.argmax(test_result[i]) == 1:
if np.argmax(test_label[i]) == 1:
TP += 1
else:
FP += 1
else:
if np.argmax(test_label[i]) == 1:
FN += 1
else:
TN += 1
precision = TP / (FP + TP)
recall = TP / (FN + TP)
print('precision', precision)
print('recall', recall)
return time_spent, precision, recall
train_dir = "/home/pdd/pdwork/CV_BiShe/picture/picTrain"
train, train_label = get_files(train_dir, 224, 224)
import numpy as np
import time
from keras.models import load_model
from keras.utils.generic_utils import CustomObjectScope
import keras
from util.dataset import get_files
test_dir = '/home/pdd/pdwork/CV_BiShe/picture/trainTest/'
test, test_label = get_files(test_dir, 224, 224)
def test_model(test, test_label, model_name):
#test_dir= '/home/pdd/pdwork/CV_BiShe/picture/picValidation/'
TP = 0
TN = 0
FP = 0
FN = 0
if (model_name[-3:] != '.h5'):
model_name += '.h5'
with CustomObjectScope({
'relu6':
keras.applications.mobilenet.relu6,
'DepthwiseConv2D':
keras.applications.mobilenet.DepthwiseConv2D
}):
model = load_model(model_name) #('DenseDWNet.h5')
time_start = time.time()
test_result = model.predict(test)
time_end = time.time()
test_result = model.predict(test)
time_end = time.time()
time_spent = time_end - time_start
print(time_spent)
#calc precision and recall
for i in range(0, len(test_label)):
if np.argmax(test_result[i]) == 1:
if np.argmax(test_label[i]) == 1:
TP += 1
else:
FP += 1
else:
if np.argmax(test_label[i]) == 1:
FN += 1
else:
TN += 1
print(TP, TN, FP, FN)
precision = TP / (FP + TP)
recall = TP / (FN + TP)
print('precision', precision)
print('recall', recall)
return time_spent, precision, recall
#train_dir ="/home/pdd/pdwork/CV_BiShe/picture/picTrain"
train_dir = "/home/pdd/pdwork/CV_BiShe/picture/trainTest"
train, train_label = get_files(train_dir, 227, 227)
train_alexnet(train, train_label)
from model.mobilenet_v2 import MobileNetv2
from model.proposed import DenseDWNet
import matplotlib.pyplot as plt
from util.dataset import get_files
from keras.callbacks import ReduceLROnPlateau, CSVLogger, TensorBoard, ModelCheckpoint, LearningRateScheduler
from keras.preprocessing.image import ImageDataGenerator, array_to_img, img_to_array, load_img
train_dir = "/home/pdd/pdwork/CV_BiShe/picture/picTrain"
#train_dir ="/home/pdd/pdwork/CV_BiShe/picture/picTrain"
x_train, y_train = get_files(train_dir, 224, 224)
validation_dir = "/home/pdd/pdwork/CV_BiShe/picture/picValidation"
x_test, y_test = get_files(train_dir, 224, 224)
#model = MobileNetv2((224, 224, 3), 2)
model = DenseDWNet((224, 224, 3), 2)
model.compile(loss="categorical_crossentropy",
optimizer='sgd',
metrics=['accuracy'])
log = CSVLogger('result/log_epoch.csv', separator=',', append=False)
board = TensorBoard(log_dir='result/logs',
histogram_freq=0,
write_graph=True,
write_images=False,
embeddings_freq=0,
embeddings_layer_names=None,
embeddings_metadata=None)
filepath = "model_{epoch:02d}-{val_acc:.2f}.h5"
checkpoint = ModelCheckpoint(filepath,
monitor='val_acc',
