自动编码器输入形状：预期 input_1 的形状为 (256, 256, 3) 但得到的数组形状为 (256, 256, 4)

Question

I am trying to build an auto-encoder but I get the following error and I can't why.

ValueError: Error when checking input: expected input_1 to have shape (256, 256, 3) but got array with shape (256, 256, 4)

If I print image shape I get (256, 256, 3) but I still get an error regarding shape.

Any help would be fantastic.

Ubuntu 18.04 | Python 3.7.6 | Tensorflow 2.1

---

from tensorflow.keras.layers import Input, Dense, Conv2D, MaxPooling2D, Dropout, Conv2DTranspose, UpSampling2D, add
from tensorflow.keras.models import Model
from tensorflow.keras import regularizers

import os
import re
from scipy import ndimage, misc
from skimage.transform import resize, rescale
from matplotlib import pyplot
import numpy as np

#Functions

def train_batches(just_load_dataset=False):

    batches = 256 # Number of images to have at the same time in a batch

    batch = 0 # Number if images in the current batch (grows over time and then resets for each batch)
    batch_nb = 0 # Batch current index

    max_batches = -1 # If you want to train only on a limited number of images to finish the training even faster.

    ep = 4 # Number of epochs

    images = []
    x_train_n = []
    x_train_down = []

    x_train_n2 = [] # Resulting high res dataset
    x_train_down2 = [] # Resulting low res dataset

    for root, dirnames, filenames in os.walk(input_dir):
        for filename in filenames:
            if re.search("\.(jpg|jpeg|JPEG|png|bmp|tiff)$", filename):
                if batch_nb == max_batches: # If we limit the number of batches, just return earlier
                    return x_train_n2, x_train_down2
                filepath = os.path.join(root, filename)
                image = pyplot.imread(filepath)

                if len(image.shape) > 2:

                    image_resized = resize(image, (256, 256))
                    x_train_n.append(image_resized)
                    x_train_down.append(rescale(rescale(image_resized, 0.5), 2.0))
                    batch += 1
                    if batch == batches:
                        batch_nb += 1

                        x_train_n2 = np.array(x_train_n)
                        x_train_down2 = np.array(x_train_down)

                        if just_load_dataset:
                            return x_train_n2, x_train_down2

                        print('Training batch', batch_nb, '(', batches, ')')

                        autoencoder.fit(x_train_down2, x_train_n2, epochs=ep, batch_size=10, shuffle=True, validation_split=0.15)

                        x_train_n = []
                        x_train_down = []

                        batch = 0

    return x_train_n2, x_train_down2

#Script

input_dir="/mnt/vanguard/datasets/ffhq-dataset/thumbnails256x256"

n = 256
chan = 3
input_img = Input(shape=(n, n, chan))


# Encoder
l1 = Conv2D(64, (3, 3), padding='same', activation='relu', activity_regularizer=regularizers.l1(10e-10))(input_img)
l2 = Conv2D(64, (3, 3), padding='same', activation='relu', activity_regularizer=regularizers.l1(10e-10))(l1)
l3 = MaxPooling2D(padding='same')(l2)
l3 = Dropout(0.3)(l3)
l4 = Conv2D(128, (3, 3),  padding='same', activation='relu', activity_regularizer=regularizers.l1(10e-10))(l3)
l5 = Conv2D(128, (3, 3), padding='same', activation='relu', activity_regularizer=regularizers.l1(10e-10))(l4)
l6 = MaxPooling2D(padding='same')(l5)
l7 = Conv2D(256, (3, 3), padding='same', activation='relu', activity_regularizer=regularizers.l1(10e-10))(l6)

# Decoder
l8 = UpSampling2D()(l7)
l9 = Conv2D(128, (3, 3), padding='same', activation='relu', activity_regularizer=regularizers.l1(10e-10))(l8)
l10 = Conv2D(128, (3, 3), padding='same', activation='relu', activity_regularizer=regularizers.l1(10e-10))(l9)
l11 = add([l5, l10])
l12 = UpSampling2D()(l11)
l13 = Conv2D(64, (3, 3), padding='same', activation='relu', activity_regularizer=regularizers.l1(10e-10))(l12)
l14 = Conv2D(64, (3, 3), padding='same', activation='relu', activity_regularizer=regularizers.l1(10e-10))(l13)
l15 = add([l14, l2])

#chan = 3, for RGB
decoded = Conv2D(chan, (3, 3), padding='same', activation='relu', activity_regularizer=regularizers.l1(10e-10))(l15)

# Create neural network
autoencoder = Model(input_img, decoded)
autoencoder_hfenn = Model(input_img, decoded)
autoencoder.summary()

autoencoder.compile(optimizer='adadelta', loss='mean_squared_error')

x_train_n = []
x_train_down = []
x_train_n, x_train_down = train_batches()

Answer 1

The rescaling x_train_down.append(rescale(rescale(image_resized, 0.5), 2.0)) is causing the problem. OpenCV can be used for degrading the image quality:

small = cv2.resize(image_resized, (0,0), fx=0.5, fy=0.5)
large = cv2.resize(small, (0,0), fx=2.0, fy=2.0)

Also, note that this is GPU-intensive computation. Either the image size should be reduced, or try GPU with more memory (K80).