valGen = HDF.HDF5DatasetGenerator(config.VAL_HDF5,
128,
preprocessors=[sp, mp, iap],
classes=2)
# 初始化优化器
print("[INFO] compiling model...")
opt = Adam(lr=1e-3)
model = alexnet.AlexNet.build(width=227,
height=227,
depth=3,
classes=2,
reg=0.0002)
model.compile(loss='binary_crossentropy', optimizer=opt, metrics=['accuracy'])
# callbacks
path = os.path.sep.join([config.OUTPUT_PATH, "{}.png".format(os.getpid())])
callbacks = [TM.TrainingMonitor(path)]
# 训练网络
model.fit_generator(trainGen.generator(),
steps_per_epoch=trainGen.numImages // 128,
validation_data=valGen.generator(),
validation_steps=valGen.numImages // 128,
epochs=75,
max_queue_size=128 * 2,
callbacks=callbacks,
verbose=1)
# 保存模型文件
print("[INFO] serializing model ....")
model.save(config.MODEL_PATH, overwrite=True)
optimizer=opt,
metrics=['accuracy'])
# 否则从磁盘中加载checkpoints模型
else:
print("[INFO] loading {}...".format(args['model']))
model = load_model(args['model'])
# 更新学习率
print("[INFO] old learning rate: {}".format(K.get_value(
model.optimizer.lr)))
K.set_value(model.optimizer.lr, 1e-5)
print("[INFO] new learning rate: {}".format(K.get_value(
model.optimizer.lr)))
# 回调函数列表
callbacks = [
# checkpoint
EPO.EpochCheckpoint(args['checkpoints'],
every=5,
startAt=args['start_epoch']),
# 监控训练过程
TM.TrainingMonitor("output/resnet56_cifar10.png",
jsonPath="output/resnet56_cifar10.json",
startAt=args['start_epoch'])
]
# 训练网络
print("[INFO] training network.....")
model.fit_generator(aug.flow(trainX, trainY, batch_size=128),
validation_data=(testX, testY),
steps_per_epoch=len(trainX) // 128,
epochs=10,
callbacks=callbacks,
verbose=1)
model.compile(loss='categorical_crossentropy', optimizer=opt, metrics=['accuracy'])
else:
print("[INFO] loading {}...".format(args["model"]))
model = load_model(args["model"])
# update the learning rate
print("[INFO] old learning rate: {}".format(K.get_value(model.optimizer.lr)))
K.set_value(model.optimizer.lr, experiment_learning_rate)
print("[INFO] new learning rate: {}".format(K.get_value(model.optimizer.lr)))
# construct the set of callbacks
callbacks = [
ECP.EpochCheckpoint(args["checkpoints"], every=5, startAt=args["start_epoch"]),
TM.TrainingMonitor(config.FIG_PATH, jsonPath=config.JSON_PATH,startAt=args["start_epoch"])]
# train the network
model.fit_generator(
trainGen.generator(),
steps_per_epoch=trainGen.numImages // 64,
validation_data=valGen.generator(),
validation_steps=valGen.numImages // 64,
epochs=experiment_epoch,
max_queue_size=64 * 2,
callbacks=callbacks,
verbose=1)
# close the databases
trainGen.close()
valGen.close()
# 初始化预处理
sp = SIP.SimplePreprocessor(64,64)
mp = MP.MeanPreprocessor(means['R'],means['G'],means['B'])
iap = ITAP.ImageToArrayPreprocessor()
# 训练数据集和验证书籍生成器
trainGen = HDFG.HDF5DatasetGenerator(config.TRAIN_HDF5,64,aug = aug,
preprocessors=[sp,mp,iap],classes=config.NUM_CLASSES)
valGen = HDFG.HDF5DatasetGenerator(config.VAL_HDF5,64,
preprocessors=[sp,mp,iap],classes=config.NUM_CLASSES)
# 回调函数列表
figPath = os.path.sep.join([args['output'],"{}.png".format(os.getpid())])
jsonPath = os.path.sep.join([args['output'],"{}.json".format(os.getpid())])
callbacks = [
TM.TrainingMonitor(figPath,jsonPath),
LearningRateScheduler(ploy_decay)
]
# 初始化模型和优化器
print("[INFO] compiling model...")
opt = SGD(lr=INIT_LR,momentum=0.9)
model = resnet.ResNet.build(32, 32, 3, config.NUM_CLASSES, (3, 4, 6), (64, 128, 256, 512), reg=0.0005,dataset="tiny_imagenet")
model.compile(loss='categorical_crossentropy', optimizer=opt, metrics=['accuracy'])
#训练网络
print("[INFO] training network....")
model.fit_generator(
trainGen.generator(),
steps_per_epoch = trainGen.numImages // 64,
validation_data = valGen.generator(),
print("[INFO] old learning rate: {}".format(K.get_value(
model.optimizer.lr)))
K.set_value(model.optimizer.lr, 1e-3)
print("[INFO] new learning rate: {}".format(K.get_value(
model.optimizer.lr)))
# construct the set of callbacks
figPath = os.path.sep.join([config.OUTPUT_PATH, "vggnet_emotion.png"])
jsonPath = os.path.sep.join([config.OUTPUT_PATH, "vggnet_emotion.json"])
callbacks = [
epochcheckpoint.EpochCheckpoint(args["checkpoints"],
every=5,
startAt=args["start_epoch"]),
trainingmonitor.TrainingMonitor(figPath,
jsonPath=jsonPath,
startAt=args["start_epoch"])
]
# train the network
model.fit_generator(trainGen.generator(),
steps_per_epoch=trainGen.numImages // config.BATCH_SIZE,
validation_data=valGen.generator(),
validation_steps=valGen.numImages // config.BATCH_SIZE,
epochs=15,
max_queue_size=config.BATCH_SIZE * 2,
callbacks=callbacks,
verbose=1)
# close the databases
trainGen.close()
testY = lb.transform(testY)
# initialize the label names for the CIFAR-10 dataset
labelNames = [
"airplane", "automobile", "bird", "cat", "deer", "dog", "frog", "horse",
"ship", "truck"
]
# initialize the SGD optimizer, but without any learning rate decay
print("[INFO] compiling model...")
opt = SGD(lr=0.01, momentum=0.9, nesterov=True)
model = minivggnet.MiniVGGNet.build(width=32, height=32, depth=3, classes=10)
model.compile(loss="categorical_crossentropy",
optimizer=opt,
metrics=["accuracy"])
# construct the set of callbacks
figPath = os.path.sep.join([args["output"], "{}.png".format(os.getpid())])
jsonPath = os.path.sep.join([args["output"], "{}.json".format(os.getpid())])
callbacks = [trainingmonitor.TrainingMonitor(figPath, jsonPath=jsonPath)]
# train the network
print("[INFO] training network...")
model.fit(trainX,
trainY,
validation_data=(testX, testY),
batch_size=64,
epochs=100,
callbacks=callbacks,
verbose=1)