此脚本演示了卷积LSTM网络的使用。

该网络用于预测包含移动方块的人工生成的电影的下一帧。

from keras.models import Sequential
from keras.layers.convolutional import Conv3D
from keras.layers.convolutional_recurrent import ConvLSTM2D
from keras.layers.normalization import BatchNormalization
import numpy as np
import pylab as plt

# 我们创建一个网络层，以尺寸为 (n_frames，width，height，channels) 的电影作为输入，并返回相同尺寸的电影。
seq = Sequential()
seq.add(ConvLSTM2D(filters=40, kernel_size=(3, 3),
                   input_shape=(None, 40, 40, 1),
                   padding='same', return_sequences=True))
seq.add(BatchNormalization())

seq.add(ConvLSTM2D(filters=40, kernel_size=(3, 3),
                   padding='same', return_sequences=True))
seq.add(BatchNormalization())

seq.add(ConvLSTM2D(filters=40, kernel_size=(3, 3),
                   padding='same', return_sequences=True))
seq.add(BatchNormalization())

seq.add(ConvLSTM2D(filters=40, kernel_size=(3, 3),
                   padding='same', return_sequences=True))
seq.add(BatchNormalization())

seq.add(Conv3D(filters=1, kernel_size=(3, 3, 3),
               activation='sigmoid',
               padding='same', data_format='channels_last'))
seq.compile(loss='binary_crossentropy', optimizer='adadelta')


# 人工数据生成：
# 生成内部有3到7个移动方块的电影。
# 方块的尺寸为 1x1 或 2x2 像素，
# 随着时间的推移线性移动。
# 为方便起见，我们首先创建宽度和高度较大的电影（80x80），最后选择 40x40 的窗口。

def generate_movies(n_samples=1200, n_frames=15):
    row = 80
    col = 80
    noisy_movies = np.zeros((n_samples, n_frames, row, col, 1), dtype=np.float)
    shifted_movies = np.zeros((n_samples, n_frames, row, col, 1),
                              dtype=np.float)

    for i in range(n_samples):
        # 添加 3 到 7 个移动方块
        n = np.random.randint(3, 8)

        for j in range(n):
            # 初始位置
            xstart = np.random.randint(20, 60)
            ystart = np.random.randint(20, 60)
            # 运动方向
            directionx = np.random.randint(0, 3) - 1
            directiony = np.random.randint(0, 3) - 1

            # 方块尺寸
            w = np.random.randint(2, 4)

            for t in range(n_frames):
                x_shift = xstart + directionx * t
                y_shift = ystart + directiony * t
                noisy_movies[i, t, x_shift - w: x_shift + w,
                             y_shift - w: y_shift + w, 0] += 1

                # 通过添加噪音使其更加健壮。
                # 这个想法是，如果在推理期间，像素的值不是一个，
                # 我们需要训练更加健壮的网络，并仍然将其视为属于方块的像素。 
                if np.random.randint(0, 2):
                    noise_f = (-1)**np.random.randint(0, 2)
                    noisy_movies[i, t,
                                 x_shift - w - 1: x_shift + w + 1,
                                 y_shift - w - 1: y_shift + w + 1,
                                 0] += noise_f * 0.1

                # Shift the ground truth by 1
                x_shift = xstart + directionx * (t + 1)
                y_shift = ystart + directiony * (t + 1)
                shifted_movies[i, t, x_shift - w: x_shift + w,
                               y_shift - w: y_shift + w, 0] += 1

    # 裁剪为 40x40 窗口
    noisy_movies = noisy_movies[::, ::, 20:60, 20:60, ::]
    shifted_movies = shifted_movies[::, ::, 20:60, 20:60, ::]
    noisy_movies[noisy_movies >= 1] = 1
    shifted_movies[shifted_movies >= 1] = 1
    return noisy_movies, shifted_movies

# 训练网络
noisy_movies, shifted_movies = generate_movies(n_samples=1200)
seq.fit(noisy_movies[:1000], shifted_movies[:1000], batch_size=10,
        epochs=300, validation_split=0.05)

# 在一部电影上测试网络
# 用前 7 个位置训练它，然后预测新的位置
which = 1004
track = noisy_movies[which][:7, ::, ::, ::]

for j in range(16):
    new_pos = seq.predict(track[np.newaxis, ::, ::, ::, ::])
    new = new_pos[::, -1, ::, ::, ::]
    track = np.concatenate((track, new), axis=0)


# 然后将预测与实际进行比较
track2 = noisy_movies[which][::, ::, ::, ::]
for i in range(15):
    fig = plt.figure(figsize=(10, 5))

    ax = fig.add_subplot(121)

    if i >= 7:
        ax.text(1, 3, 'Predictions !', fontsize=20, color='w')
    else:
        ax.text(1, 3, 'Initial trajectory', fontsize=20)

    toplot = track[i, ::, ::, 0]

    plt.imshow(toplot)
    ax = fig.add_subplot(122)
    plt.text(1, 3, 'Ground truth', fontsize=20)

    toplot = track2[i, ::, ::, 0]
    if i >= 2:
        toplot = shifted_movies[which][i - 1, ::, ::, 0]

    plt.imshow(toplot)
    plt.savefig('%i_animate.png' % (i + 1))