def predict(i, input_imgs, network, target_imgs, fo):
out0 = network.predict(input_imgs)
print(
'%dth output image, loss = %.6f, min = %.6f, max = %.6f, mean = %.6f, var = %.6f\n'
% (i, np.mean(np.abs(target_imgs[0] - out0[0])), np.min(out0[0]),
np.max(out0[0]), np.mean(out0[0]), math.sqrt(np.var(out0[0]))))
output = utils.compose_dwt_images(out0, FLAGS.wavelet)
return output
def ensem_predict(input_imgs, network):
# ensembling
outs_list = []
for _, flip_axis in enumerate([0, 1, 2, -1]):
for _, rotate_rg in enumerate([0, 90]):
en_imgs = utils.enhance_imgs(input_imgs, rotate_rg, flip_axis)
outs = network.predict(en_imgs)
composed_img = utils.compose_dwt_images(outs, FLAGS.wavelet)
anti_outs = utils.anti_enhance_imgs(composed_img, rotate_rg,
flip_axis)
outs_list.append(anti_outs[0])
output = np.mean(outs_list, axis=0)
return [output]
def predict(input_imgs, network, step):
out_hfreq, out_lowfreq = network.predict(input_imgs)
out = concat(out_hfreq, out_lowfreq, step)
output = utils.compose_dwt_images([out], FLAGS.wavelet)
return output
def predict(input_imgs, network):
out0 = network.predict(input_imgs)
output = utils.compose_dwt_images(out0, FLAGS.wavelet)
return output