ValueError: Error when checking input: expected input_2 to have shape (224, 224, 3) but got array with shape (244, 244, 3)

Question

I am trying to use a pretained CNN (VGG16) but I keep getting the following error:

ValueError: Error when checking input: expected input_2 to have shape (224, 224, 3) but got array with shape (244, 244, 3)

Here is my full code:

import numpy as np 
import keras 
from keras import backend as K 
from keras.models import Sequential 
from keras.layers import Activation 
from keras.layers.core import Dense, Flatten 
from keras.optimizers import Adam 
from keras.metrics import categorical_crossentropy 
from keras.preprocessing.image import ImageDataGenerator
from keras.layers.normalization import BatchNormalization 
from keras.layers.convolutional import *

train_path = "/DATA/train"
valid_path = "/DATA/valid"
test_path = "/DATA/test"
#creating the training, testing, and validation sets 
trainBatches = ImageDataGenerator().flow_from_directory(train_path, target_size=(244,244), classes=['classU', 'classH'], batch_size=20)
valBatches = ImageDataGenerator().flow_from_directory(valid_path, target_size=(244,244), classes=['classU', 'classH'], batch_size=2)
testBatches = ImageDataGenerator().flow_from_directory(test_path, target_size=(244,244), classes=['classU', 'classH'], batch_size=2)
#loading the model & removing the top layer 
model = Sequential() 
for layer in vgg16_model.layers[:-1]:
    model.add(layer)

#Fixing the weights 
for layer in model.layers:
    layer.trainable = False

#adding the new classier 
model.add(Dense(2, activation = 'softmax'))


model.compile(Adam(lr=0.0001), loss='categorical_crossentropy', metrics=['accuracy'])
model.fit_generator(trainBatches, steps_per_epoch=89, validation_data=valBatches, validation_steps=11, epochs=5, verbose=2)

But I don't know what I am getting the error. I thought ImageDataGenerator() will take care of the data/batches generation with the correct dimensions. What I am missing?

Answer 1

The VGG model in this case expects images to be of (224, 224) whereas your image generator targets are (244, 244) hence your input shapes get a mismatch. You should adjust the target size to the expected shape. The documentation details the expected inputs and it also has an option include_top that will remove the last layer for you so you don't have to do it manually.