model.summary()
filepath = os.path.join(data_output, "checkpoints",
"weights-improvement-{epoch:02d}-{PSNR:.2f}.hdf5")
checkpoint = ModelCheckpoint(filepath, monitor=PSNR, verbose=1, mode='max')
callbacks_list = [checkpoint]
model.fit(X, y, epochs=EPOCHS,
validation_split=0.20,
batch_size=BATCH_SIZE, callbacks=callbacks_list, shuffle="batch")
print("Done training!!!")
print("Saving the final model ...")
model.save(os.path.join(data_output,"fsrcnn_Sat-6_Epoch{}_VIS2NIR.h5".format(EPOCHS))) # creates a HDF5 file
y_pred = model.predict(x_true)
n = y_true.shape[0]
mses = 0
rmses = 0
psnrs = 0
ssims = 0
for i in range(0,n):
print("Testing {}/{}...".format(i,n))
y = y_true[i,:,:,0]
y_prime = y_pred[i,:,:,0]
mses = mses + mse(y, y_prime)
rmses = rmses + rmse(y, y_prime)
psnrs = psnrs + psnr(y, y_prime, data_range=y_prime.max() - y_prime.min())
ssims = ssims + ssim(y, y_prime, data_range=y_prime.max() - y_prime.min())
model = PReLU()(model)
model = Conv2DTranspose(1, (9, 9), strides=(4, 4), padding='same')(model)
output_img = model
model = Model(input_img, output_img)
# model.load_weights('./checkpoints/weights-improvement-20-26.93.hdf5')
model.compile(optimizer='adam',
lr=0.0001,
loss='mse',
metrics=[PSNR, "accuracy"])
model.summary()
filepath = "./checkpoints/weights-improvement-{epoch:02d}-{PSNR:.2f}.hdf5"
checkpoint = ModelCheckpoint(filepath, monitor=PSNR, verbose=1, mode='max')
callbacks_list = [checkpoint]
model.fit_generator(image_gen(train_list, scale=INPUT_SCALE), steps_per_epoch=len(train_list) // BATCH_SIZE, \
validation_data=image_gen(test_list, scale=INPUT_SCALE), validation_steps=len(train_list) // BATCH_SIZE, \
epochs=EPOCHS, workers=8, callbacks=callbacks_list)
print("Done training!!!")
print("Saving the final model ...")
model.save('fsrcnn_model.h5') # creates a HDF5 file
del model # deletes the existing model
checkpoint = ModelCheckpoint(filepath, monitor=PSNR, verbose=1, mode='max')
callbacks_list = [checkpoint]
model.fit(X,
y,
epochs=EPOCHS,
validation_split=0.25,
batch_size=BATCH_SIZE,
callbacks=callbacks_list,
shuffle="batch")
print("Done training!!!")
print("Saving the final model ...")
model.save(
os.path.join(data_output, "fsrcnn_L7_Epoch{}_{}x.h5".format(
EPOCHS, SCALE_FACTOR))) # creates a HDF5 file
y_pred = model.predict(x_true)
n = y_true.shape[0]
mses = 0
rmses = 0
psnrs = 0
ssims = 0
for i in range(0, n):
print("Testing {}/{}...".format(i, n))
y = y_true[i, :, :, 0]
y_prime = y_pred[i, :, :, 0]
mses = mses + mse(y, y_prime)
rmses = rmses + rmse(y, y_prime)
psnrs = psnrs + psnr(y, y_prime, data_range=y_prime.max() - y_prime.min())
