正文共5042个字，11张图，预计阅读时间13分钟。

01

Alpha版本：40行代码，实现基础着色机器人

from keras.layers import Conv2D, UpSampling2D, InputLayer, Conv2DTranspose

from keras.layers import Activation, Dense, Dropout, Flatten

from keras.layers.normalization import BatchNormalization

from keras.models import Sequential

from keras.preprocessing.image

import ImageDataGenerator, array_to_img, img_to_array, load_img

from skimage.color import rgb2lab, lab2rgb, rgb2gray, xyz2lab

from skimage.io import imsave

import numpy as np

import os

import random

import tensorflow as tf

Using TensorFlow backend.# Get

imagesimage = img_to_array(load_img('woman.jpg')) image = np.array(image, dtype=float) X = rgb2lab(1.0/255*image)[:,:,0] Y = rgb2lab(1.0/255*image)[:,:,1:] Y /= 128X = X.reshape(1, 400, 400, 1) Y = Y.reshape(1, 400, 400, 2)# Building the neural networkmodel = Sequential() model.add(InputLayer(input_shape=(None, None, 1))) model.add(Conv2D(8, (3, 3), activation='relu', padding='same', strides=2)) model.add(Conv2D(8, (3, 3), activation='relu', padding='same')) model.add(Conv2D(16, (3, 3), activation='relu', padding='same')) model.add(Conv2D(16, (3, 3), activation='relu', padding='same', strides=2)) model.add(Conv2D(32, (3, 3), activation='relu', padding='same')) model.add(Conv2D(32, (3, 3), activation='relu', padding='same', strides=2)) model.add(UpSampling2D((2, 2))) model.add(Conv2D(32, (3, 3), activation='relu', padding='same')) model.add(UpSampling2D((2, 2))) model.add(Conv2D(16, (3, 3), activation='relu', padding='same')) model.add(UpSampling2D((2, 2))) model.add(Conv2D(2, (3, 3), activation='tanh', padding='same'))# Finish modelmodel.compile(optimizer='rmsprop', loss='mse') model.fit(x=X, y=Y, batch_size=1, epochs=1000)

Epoch 1/1000

1/1 [==============================] - 1s - loss: 0.0286 Epoch 2/1000 1/1 [==============================] - 0s - loss: 0.0238 Epoch 318/1000 1/1 [==============================] - 0s - loss: 0.0010 Epoch 319/1000 1/1 [==============================] - 0s - loss: 7.4259e-04 Epoch 590/1000 1/1 [==============================] - 0s - loss: 5.5838e-04 Epoch 591/1000 1/1 [==============================] - 0s - loss: 4.7110e-04 Epoch 592/1000 Epoch 845/1000 1/1 [==============================] - 0s - loss: 3.5430e-04 Epoch 846/1000 1/1 [==============================] - 0s - loss: 2.9861e-04 1/1 [==============================] - 0s - loss: 3.0116e-04 Epoch 996/1000 1/1 [==============================] - 0s - loss: 3.1555e-04 Epoch 997/1000 1/1 [==============================] - 0s - loss: 3.0418e-04 Epoch 998/1000 1/1 [==============================] - 0s - loss: 4.3305e-04 Epoch 999/1000 1/1 [==============================] - 0s - loss: 3.9781e-04 Epoch 1000/1000 1/1 [==============================] - 0s - loss: 5.8701e-04<keras.callbacks.History at 0x11ccb6860>

print(model.evaluate(X, Y, batch_size=1)) output = model.predict(X) output *= 128# Output colorizationscur = np.zeros((400, 400, 3)) cur[:,:,0] = X[0][:,:,0] cur[:,:,1:] = output[0] imsave("img_result.png", lab2rgb(cur)) imsave("img_gray_version.png", rgb2gray(lab2rgb(cur)))

1/1 [==============================] - 0s 0.000459772680188 /usr/local/lib/python3.6/site-packages/skimage/util/dtype.py:122: UserWarning: Possible precision loss when converting from float64 to uint8

.format(dtypeobj_in, dtypeobj_out))

/usr/local/lib/python3.6/site-packages/skimage/util/dtype.py:122: UserWarning: Possible precision loss when converting from float64 to uint16  .format(dtypeobj_in, dtypeobj_out))

# 可视化数据集

import matplotlib.pyplot as plt %matplotlib inline   plt.rcParams['font.sans-serif']=['SimHei'] #用来正常显示中文标签

img = lab2rgb(cur) title = '黑白照片自动着色的神经网络-Alpha版'plt.imshow(img) plt.title(title)

plt.show()