mnist CNN ValueError expected min_ndim=4, found ndim=3. Full shape received: [32, 28, 28]
Question:
I define the model definition as follows.
tf.keras.datasets.mnist
model = keras.models.Sequential([
tf.keras.layers.Conv2D(28, (3,3), activation='relu', input_shape=(28, 28, 1)),
tf.keras.layers.MaxPooling2D((2, 2)),
tf.keras.layers.Conv2D(56, (3,3), activation='relu'),
tf.keras.layers.Flatten(),
tf.keras.layers.Dense(64, activation='relu'),
tf.keras.layers.Dense(10, activation='softmax'),
])
model.fit(x_train, y_train, epochs=3) <----error
When I try to run with my data set, it gives the following error.
ValueError: Input 0 of layer sequential_3 is incompatible with the layer: : expected min_ndim=4, found ndim=3. Full shape received: [32, 28, 28]
Answers:
By seeing your error, I think you probably didn’t add the channel axis in the training set i.e [batch, w, h, channel
]. Here is the working code
DataSet
import tensorflow as tf
import numpy as np
from sklearn.model_selection import train_test_split
(x_train, y_train), (x_test, y_test) = tf.keras.datasets.mnist.load_data()
x_train = np.expand_dims(x_train, axis=-1) # <--- add channel axis
x_train = x_train.astype('float32') / 255
y_train = tf.keras.utils.to_categorical(y_train, num_classes=10)
print(x_train.shape, y_train.shape)
# (60000, 28, 28, 1) (60000, 10)
Training
model = tf.keras.models.Sequential([
tf.keras.layers.Conv2D(28, (3,3), activation='relu', input_shape=(28, 28, 1)),
tf.keras.layers.MaxPooling2D((2, 2)),
tf.keras.layers.Conv2D(56, (3,3), activation='relu'),
tf.keras.layers.Flatten(),
tf.keras.layers.Dense(64, activation='relu'),
tf.keras.layers.Dense(10, activation='softmax'),
])
model.compile(loss='categorical_crossentropy',optimizer='Adam',metrics=['accuracy'])
model.fit(x_train, y_train, epochs=3)
Epoch 1/3
1875/1875 [==============================] - 11s 2ms/step - loss: 0.2803 - accuracy: 0.9160
Epoch 2/3
1875/1875 [==============================] - 4s 2ms/step - loss: 0.0434 - accuracy: 0.9869
Epoch 3/3
1875/1875 [==============================] - 4s 2ms/step - loss: 0.0273 - accuracy: 0.9917
another way to solve it:
x_train, x_test = x_train.reshape(-1,28,28,1), x_test.reshape(-1,28,28,1)
I define the model definition as follows.
tf.keras.datasets.mnist
model = keras.models.Sequential([
tf.keras.layers.Conv2D(28, (3,3), activation='relu', input_shape=(28, 28, 1)),
tf.keras.layers.MaxPooling2D((2, 2)),
tf.keras.layers.Conv2D(56, (3,3), activation='relu'),
tf.keras.layers.Flatten(),
tf.keras.layers.Dense(64, activation='relu'),
tf.keras.layers.Dense(10, activation='softmax'),
])
model.fit(x_train, y_train, epochs=3) <----error
When I try to run with my data set, it gives the following error.
ValueError: Input 0 of layer sequential_3 is incompatible with the layer: : expected min_ndim=4, found ndim=3. Full shape received: [32, 28, 28]
By seeing your error, I think you probably didn’t add the channel axis in the training set i.e [batch, w, h, channel
]. Here is the working code
DataSet
import tensorflow as tf
import numpy as np
from sklearn.model_selection import train_test_split
(x_train, y_train), (x_test, y_test) = tf.keras.datasets.mnist.load_data()
x_train = np.expand_dims(x_train, axis=-1) # <--- add channel axis
x_train = x_train.astype('float32') / 255
y_train = tf.keras.utils.to_categorical(y_train, num_classes=10)
print(x_train.shape, y_train.shape)
# (60000, 28, 28, 1) (60000, 10)
Training
model = tf.keras.models.Sequential([
tf.keras.layers.Conv2D(28, (3,3), activation='relu', input_shape=(28, 28, 1)),
tf.keras.layers.MaxPooling2D((2, 2)),
tf.keras.layers.Conv2D(56, (3,3), activation='relu'),
tf.keras.layers.Flatten(),
tf.keras.layers.Dense(64, activation='relu'),
tf.keras.layers.Dense(10, activation='softmax'),
])
model.compile(loss='categorical_crossentropy',optimizer='Adam',metrics=['accuracy'])
model.fit(x_train, y_train, epochs=3)
Epoch 1/3
1875/1875 [==============================] - 11s 2ms/step - loss: 0.2803 - accuracy: 0.9160
Epoch 2/3
1875/1875 [==============================] - 4s 2ms/step - loss: 0.0434 - accuracy: 0.9869
Epoch 3/3
1875/1875 [==============================] - 4s 2ms/step - loss: 0.0273 - accuracy: 0.9917
another way to solve it:
x_train, x_test = x_train.reshape(-1,28,28,1), x_test.reshape(-1,28,28,1)