输出预测图像并存起来
'''
from model import RDN
import skimage.io
import glob
import os
import numpy as np
from train import modelsavedir
if __name__ == '__main__':
#xdir='../Data/Set5_test_LR'
xdir = '../Data/Urban_test_LR'
xlist = glob.glob(os.path.join(xdir, '*.png'))
myRDN = RDN()
#save_dir='./result'
save_dir = './result2'
modellist = glob.glob(modelsavedir + '*RDN*.hdf5')
if len(modellist) > 0:
modellist.sort(key=lambda x: float(x[len(modelsavedir) + 9:len(
modelsavedir) + 13]))
model = myRDN.load_weight(modellist[0])
print('载入', modellist[0])
else:
model = myRDN.load_weight()
# 读取图像
for imgname in xlist:
print(imgname)
img = skimage.io.imread(imgname)
Y = model.predict(np.array([img]), 1)[0]
Y = np.clip(Y, 0, 255)
Y = Y.astype(np.uint8)
skimage.io.imsave(os.path.join(save_dir, os.path.basename(imgname)), Y)
modelcp = ModelCheckpoint(modelsavedir +
'{epoch:04d}-RDN-{val_loss:.2f}-weights.hdf5',
verbose=1,
period=1,
save_weights_only=True,
save_best_only=True)
gen_tx, gen_ty, gen_vx, gen_vy = get_train_data(Batch_size)
train_gen = gen(gen_tx, gen_ty)
valid_gen = gen(gen_vx, gen_vy)
myrdn.setting_train()
# 载入之前的模型
modellist = glob.glob(modelsavedir + '*RDN*.hdf5')
modellist.sort(
key=lambda x: float(x[len(modelsavedir) + 0:len(modelsavedir) + 4]))
myrdn.load_weight(modellist[-1])
print('载入', modellist[-1])
init_epoch = int(os.path.basename(modellist[-1])[:4])
target_epoch = 170
step = 400
print('目标', target_epoch, '还有', target_epoch - init_epoch, '\n时间（分）',
3 / (500 * 24) * step * Batch_size * (target_epoch - init_epoch))
try:
myrdn.model.fit_generator(train_gen,
step,
epochs=target_epoch,
verbose=1,
validation_data=valid_gen,
validation_steps=6,
callbacks=[lr_decay, modelcp],