Question

我想使用 convLSTM 进行降水预测我正在使用来自 github

的代码

该代码是为多 GPU 设计的。

我使用过：tf.config.experimental.list_physical_devices('GPU')，因为我只有一个 GPU

import numpy as np
from PIL import Image
import os
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 matplotlib.pyplot as plt
import time
from keras.utils import multi_gpu_model
from keras import optimizers
import tensorflow as tf

WIDTH = 646
HEIGHT = 631
FRAMES = 0

SEQUENCE = np.load('D:/IMG2/sequence_array.npz')['sequence_array']  # load array
print(SEQUENCE[0])
print('Data loaded.')
print(time.strftime('%Y-%m-%d %H:%M:%S', time.localtime()))

NUMBER = len(SEQUENCE)


i = 0
while i < NUMBER:
if (i + 1) % 11 != 0:
    BASIC_SEQUENCE = np.append(BASIC_SEQUENCE, SEQUENCE[i])
    NEXT_SEQUENCE = np.append(NEXT_SEQUENCE, SEQUENCE[i+1])
i += 1
print(i)
print(time.strftime('%Y-%m-%d %H:%M:%S', time.localtime()))

SEQUENCE = SEQUENCE.reshape(NUMBER, WIDTH, HEIGHT, 3)
BASIC_SEQUENCE = np.zeros((NUMBER-FRAMES, FRAMES, WIDTH, HEIGHT, 3))
NEXT_SEQUENCE = np.zeros((NUMBER-FRAMES+1, FRAMES+1, WIDTH, HEIGHT, 3))

# step =1
SEQUENCE = SEQUENCE.reshape(NUMBER, WIDTH, HEIGHT, 3)
# step =2
SEQUENCE_2 = []
for i in range(int(NUMBER / 2)):
SEQUENCE_2.append(SEQUENCE[2 * i])

# step = 3
SEQUENCE_3 = []
for i in range(int(NUMBER / 3)):
SEQUENCE_3.append(SEQUENCE[3 * i])

for i in range(FRAMES):
print(i)
BASIC_SEQUENCE[:, i, :, :, :] = SEQUENCE[i:i+NUMBER-FRAMES]
NEXT_SEQUENCE[:, i, :, :, :] = SEQUENCE[i+1:i+NUMBER-FRAMES+1]



#plt.imshow(BASIC_SEQUENCE[0][0].reshape(100, 100))
#plt.show()
# build model


seq = Sequential()

seq.add(ConvLSTM2D(filters=40, kernel_size=(3, 3),input_shape=(None, WIDTH, HEIGHT, 3), padding='same', return_sequences=True))
seq.add(BatchNormalization())

seq.add(ConvLSTM2D(filters=60, kernel_size=(3, 3), padding='same', return_sequences=True))
seq.add(BatchNormalization())

seq.add(ConvLSTM2D(filters=60, 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='mean_squared_error', optimizer='adadelta')

seq.fit(BASIC_SEQUENCE[:10], NEXT_SEQUENCE[:10], batch_size=32,
    epochs=2, validation_split=0.05)


parallel_model = tf.config.experimental.list_physical_devices('GPU')
sgd = optimizers.SGD(lr=0.01, clipnorm=1)

loss='mean_squared_error'
optimizer='adadelta'

BASIC_SEQUENCE[:1000]
NEXT_SEQUENCE[:1000]
batch_size=10
epochs=10
validation_split=0.05


seq.save('nice_model.h5')

track = BASIC_SEQUENCE[::, ::, ::, ::]

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

# And then compare the predictions
# to the ground truth
track2 = BASIC_SEQUENCE[::, ::, ::, ::]
for i in range(FRAMES):
fig = plt.figure(figsize=(10, 5))
ax = fig.add_subplot(121)
if i >= 8:
    ax.text(1, 3, 'Predictions !', fontsize=20, color='w')
else:
    ax.text(1, 3, 'Inital trajectory', fontsize=20)
toplot = track[i, ::, ::, 0]
plt.imshow(toplot, cmap='binary')
ax = fig.add_subplot(122)
plt.text(1, 3, 'Ground truth', fontsize=20)
toplot = track2[i, ::, ::, 0]
if i >= 8:
    #toplot = NEXT_SEQUENCE[which][i - 1, ::, ::, 0]
    toplot = NEXT_SEQUENCE[i - 1, ::, ::, 0]
plt.imshow(toplot, cmap='binary')
plt.savefig('%i_animate.png' % (i + 1))

但是，我有尺寸问题。

我该如何解决这些问题？

ValueError:Input 0 of layer convlstm2d_46 is incompatible with the layer:expected ndim=5,found ndim=6.Full shape received: [None, 100, 0, 646, 631, 3]

0 个答案: