ValueError: Input 0 of layer "model_10" is incompatible with the layer: expected shape=(None, 244, 244, 3), found shape=(None, 224, 224, 3)

Question

I am training a model on top of the prebuilt imagenetV2 model to classify dog breeds.

Here is my code.

import os
import tensorflow as tf

\_URL = 'http://vision.stanford.edu/aditya86/ImageNetDogs/images.tar'
path_to_zip = tf.keras.utils.get_file('images.tar', origin=\_URL, extract=True)

BATCH_SIZE = 32
IMG_SIZE = (224, 224)
dir = os.path.join(os.path.dirname(path_to_zip), 'Images')
train_dataset = tf.keras.utils.image_dataset_from_directory(dir,
shuffle=True,
batch_size=BATCH_SIZE,
image_size=IMG_SIZE,
validation_split=.2,
subset='training',
seed=2021)

validation_dataset = tf.keras.utils.image_dataset_from_directory(dir,
shuffle=True,
batch_size=BATCH_SIZE,
image_size=IMG_SIZE,
validation_split=.2,
subset='validation',
seed=2021)

len(train_dataset.class_names)

import matplotlib.pyplot as plt
class_names = train_dataset.class_names

plt.figure(figsize=(10, 10))
for images, labels in train_dataset.take(1):
for i in range(9):
ax = plt.subplot(3, 3, i + 1)
plt.imshow(images\[i\].numpy().astype("uint8"))
plt.title(class_names\[labels\[i\]\])
plt.axis("off")

AUTOTUNE = tf.data.AUTOTUNE

train_dataset = train_dataset.prefetch(buffer_size=AUTOTUNE)
validation_dataset = validation_dataset.prefetch(buffer_size=AUTOTUNE)

val_batches = tf.data.experimental.cardinality(validation_dataset)

data_augmentation = tf.keras.Sequential(\[
tf.keras.layers.RandomFlip('horizontal'),
tf.keras.layers.RandomRotation(0.2),
\])

for image, \_ in train_dataset.take(1):
plt.figure(figsize=(10, 10))
first_image = image\[0\]
for i in range(9):
ax = plt.subplot(3, 3, i + 1)
augmented_image = data_augmentation(tf.expand_dims(first_image, 0))
plt.imshow(augmented_image\[0\] / 255)
plt.axis('off')

rescale = tf.keras.layers.Rescaling(1./127.5, offset=-1)

IMG_SHAPE = IMG_SIZE + (3,)
base_model = tf.keras.applications.MobileNetV2(IMG_SHAPE,
include_top=False,
weights='imagenet')

base_model.trainable = False

global_average_layer = tf.keras.layers.GlobalAveragePooling2D()

model = tf.keras.Sequential()
model.add(tf.keras.Input(shape=(244, 244, 3, )))
model.add(tf.keras.layers.RandomFlip('horizontal'))
model.add(tf.keras.layers.RandomRotation(0.2))
model.add(rescale)
model.add(base_model)
model.add(global_average_layer)
model.add(tf.keras.layers.Dropout(0.2))
model.add(tf.keras.layers.Dense(120))
model.summary()

At this point I get the following warning

WARNING:tensorflow:Model was constructed with shape (None, 224, 224, 3) for input KerasTensor(type_spec=TensorSpec(shape=(None, 224, 224, 3), dtype=tf.float32, name='input_1'), name='input_1', description="created by layer 'input_1'"), but it was called on an input with incompatible shape (None, 244, 244, 3).

And after I compile the model like below

model.compile(optimizer=tf.keras.optimizers.Adam(learning_rate=base_learning_rate),
loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=False),
metrics=\['accuracy'\])

model.summary()

initial_epochs = 20

I get no errors but when I evaluate like so

loss0, accuracy0 = model.evaluate(train_dataset)

I get the following stack trace

ValueError: in user code:

    File "/usr/local/lib/python3.7/dist-packages/keras/engine/training.py", line 1557, in test_function  *
        return step_function(self, iterator)
    File "/usr/local/lib/python3.7/dist-packages/keras/engine/training.py", line 1546, in step_function  **
        outputs = model.distribute_strategy.run(run_step, args=(data,))
    File "/usr/local/lib/python3.7/dist-packages/keras/engine/training.py", line 1535, in run_step  **
        outputs = model.test_step(data)
    File "/usr/local/lib/python3.7/dist-packages/keras/engine/training.py", line 1499, in test_step
        y_pred = self(x, training=False)
    File "/usr/local/lib/python3.7/dist-packages/keras/utils/traceback_utils.py", line 67, in error_handler
        raise e.with_traceback(filtered_tb) from None
    File "/usr/local/lib/python3.7/dist-packages/keras/engine/input_spec.py", line 264, in assert_input_compatibility
        raise ValueError(f'Input {input_index} of layer "{layer_name}" is '
    
    ValueError: Input 0 of layer "model_10" is incompatible with the layer: expected shape=(None, 244, 244, 3), found shape=(None, 224, 224, 3)

I have double checked the shape of everything and I am not sure what is causing this issue. I also have checked that the seed includes all classes for both training and validation. I suspect it has something to do with the transferred imagenet model but I have followed tensorflow tutorials closely and I have not seen a problem with the way I am doing things. Any and all help is appreciated thank you

Answer 1

The error is due to the shape mismatch. Your input image is of shape (224, 224, 3) but the shape in the input layer is (244, 244, 3). Both the shapes should be same.

model.add(tf.keras.Input(shape=(224, 224, 3)))

Kindly change the input shape as above to avoid the error. Thank you!