((trainX, trainY), (testX, testY)) = cifar10.load_data()
trainX = trainX.astype("float") / 255.0
testX = testX.astype("float") / 255.0
lb = LabelBinarizer()
trainY = lb.fit_transform(trainY)
testY = lb.transform(testY)
# 标签0-9代表的类别string
labelNames = [
'airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse',
'ship', 'truck'
]
print("[INFO] compiling model...")
opt = SGD(lr=0.01, decay=0.01 / 40, momentum=0.9, nesterov=True)
model = MiniVGGNet.build(width=32, height=32, depth=3, classes=10)
model.compile(loss="categorical_crossentropy",
optimizer=opt,
metrics=["accuracy"])
# 定义模型保存的模板
fname = os.path.sep.join(
[args["weights"], "weights={epoch:03d}-{val_loss:.4f}.hdf5"])
# 调用keras的模块然后根据参数进行保存，具体的monitor mode等参数我觉得最后的学习方法就是直接去看原文档
checkpoint = ModelCheckpoint(fname,
monitor="val_loss",
mode="min",
save_best_only=True,
verbose=1)
callbacks = [checkpoint]
print("[INFO] training our network MiniVGG...")
data = data.reshape(data.shape[0], 28, 28, 1)
(trainX, testX, trainY, testY) = train_test_split(data / 255.0,
dataset.target.astype("int"),
test_size=0.25,
random_state=42)
# 将label进行one-hot编码
lb = LabelBinarizer()
trainY = lb.fit_transform(trainY)
testY = lb.transform(testY)
print("[INFO] compiling model...")
callbacks = [LearningRateScheduler(step_decay)]
# opt = SGD(lr=0.01, decay=0.01 / 70, momentum=0.9, nesterov=True)
opt = SGD(lr=0.01, momentum=0.9, nesterov=True)
model = MiniVGGNet.build(width=28, height=28, depth=1, classes=10)
model.compile(loss="categorical_crossentropy",
optimizer=opt,
metrics=['accuracy'])
print(model.summary())
print("[INFO] training network miniVGG...")
H = model.fit(trainX,
trainY,
validation_data=(testX, testY),
batch_size=64,
epochs=70,
callbacks=callbacks,
verbose=1)
# H = model.fit(trainX, trainY, validation_data=(testX, testY), batch_size=64, epochs=70, verbose=1)
model.save(args["model"])
lr = 0.01
bs = 32
checkpoint_filepath = "./train/best_model.h5"
log_dir = f"./logs/fit/miniVGGNET2_data_augm_weighted_{lr}_bs{bs}"
callbacks = [
tf.keras.callbacks.TensorBoard(log_dir=log_dir, histogram_freq=1),
tf.keras.callbacks.LearningRateScheduler(stepwise_scheduler, verbose=1),
tf.keras.callbacks.ModelCheckpoint(filepath=checkpoint_filepath,
monitor="val_loss",
verbose=1,
save_best_only=True),
]
model = MiniVGGNet.build(width, height, depth, 2)
# compile
optimizer = tf.keras.optimizers.SGD(learning_rate=lr)
model.compile(optimizer=optimizer,
loss="binary_crossentropy",
metrics=["accuracy"])
# fit with data augm
h = model.fit(data_augm.flow(x=X_train, y=y_train, batch_size=bs),
steps_per_epoch=len(X_train) // bs,
epochs=epochs,
verbose=1,
validation_data=(X_test, y_test),
shuffle=True,
callbacks=callbacks,
class_weight=classes_weight)
trainX = data.train.images
trainY = data.train.labels
testX = data.valid.images
testY = data.vaild.labels
# ((trainX, trainY), (testX, testY)) = cifar10.load_data()
# trainX = trainX.astype("float") / 255.0
# testX = testX.astype("float") / 255.0
lb = LabelBinarizer()
trainY = lb.fit_transform(trainY)
testY = lb.transform(testY)
print("[INFO] compiling model...")
opt = SGD(lr=0.01, decay=0.01 / 70, momentum=0.9, nesterov=True)
model = MiniVGGNet.build(width=48, height=48, depth=2, classes=2)
model.compile(loss="categorical_crossentropy",
optimizer=opt,
metrics=['accuracy'])
print(model.summary())
print("[INFO] training network Lenet-5")
H = model.fit(trainX,
trainY,
validation_data=(testX, testY),
batch_size=64,
epochs=70,
callbacks=callbacks,
verbose=1)
model.save(args["model"])
from lenet import LeNet from miniVGG import MiniVGGNet from keras.utils import plot_model model1 = LeNet.build(28, 28, 1, 10) plot_model(model1, to_file="lenet-5-mnist.png", show_shapes=True) model12 = LeNet.build(32, 32, 3, 10) plot_model(model12, to_file="lenet-5-cifar10.png", show_shapes=True) model13 = LeNet.build(32, 32, 3, 100) plot_model(model13, to_file="lenet-5-cifar100.png", show_shapes=True) model2 = MiniVGGNet.build(28, 28, 1, 10) plot_model(model2, to_file="miniVGGNet-mnist.png", show_shapes=True) model21 = MiniVGGNet.build(32, 32, 3, 10) plot_model(model21, to_file="miniVGGNet-cifar10.png", show_shapes=True) model22 = MiniVGGNet.build(32, 32, 3, 100) plot_model(model22, to_file="miniVGGNet-cifar100.png", show_shapes=True)
sdl = SimpleDatasetLoader(preprocessors=[aap, iap])
(data, labels) = sdl.load(imagePaths, verbose=500)
data = data.astype("float") / 255.0
(trainX, testX, trainY, testY) = train_test_split(data, labels, test_size=0.25, random_state=42)
trainY = LabelBinarizer().fit_transform(trainY)
testY = LabelBinarizer().fit_transform(testY)
aug = ImageDataGenerator(rotation_range=40, width_shift_range=0.1, height_shift_range=0.1,
shear_range=0.2, zoom_range=0.2, horizontal_flip=True, fill_mode="nearest")
print("[INFO] compiling model...")
# opt = SGD(lr=0.05)
opt = SGD(lr=0.05, decay=0.05 / 200, momentum=0.9, nesterov=True)
model = MiniVGGNet.build(width=64, height=64, depth=3, classes=len(classNames))
model.compile(loss="categorical_crossentropy", optimizer=opt, metrics=["accuracy"])
checkpoint = ModelCheckpoint(args["weights"], monitor="val_loss", mode="min", save_best_only=True, verbose=1)
callbacks = [checkpoint]
print("[INFO] training network...")
# H = model.fit(trainX, trainY, validation_data=(testX, testY), batch_size=32, epochs=100, verbose=1)
H = model.fit_generator(aug.flow(trainX, trainY, batch_size=32), validation_data=(testX, testY),
steps_per_epoch=len(trainX) // 32, epochs=200, verbose=1)
print("[INFO] evaluating network...")
preds = model.predict(testX, batch_size=32)
print(classification_report(testY.argmax(axis=1), preds.argmax(axis=1), target_names=classNames))
# 保存可视化训练结果